HK1243571B - Methods and devices for video coding - Google Patents

Description

Method and apparatus for video decoding

Information about divisional applications

This application is a divisional application. The parent application is an international application filed on September 20, 2012, with application number PCT/US2012/056362, which, after entering China, is titled “Reference Picture List Construction for Video Decoding” and has application number 201280046488.0.

This application claims the rights of:

U.S. Provisional Application No. 61/538,787, filed September 23, 2011;

U.S. Provisional Patent Application No. 61/539,433, filed September 26, 2011; and

U.S. Provisional Patent Application No. 61/542,034, filed September 30, 2011, the entire contents of each of which are incorporated herein by reference in their entirety.

Technical Field

This disclosure relates to video coding and, more particularly, to techniques for coding video data.

Background Art

Digital video capabilities can be incorporated into a wide range of devices, including digital televisions, digital live broadcast systems, wireless broadcast systems, personal digital assistants (PDAs), laptop or desktop computers, tablet computers, e-book readers, digital camcorders, digital recording devices, digital media players, video game devices, video game consoles, cellular or satellite radio telephones, so-called "smartphones," video teleconferencing devices, video streaming devices, and the like. Digital video devices implement video compression techniques, such as those described in the standards defined by MPEG-2, MPEG-4, ITU-T H.263, ITU-T H.264/MPEG-4 Part 10 (Advanced Video Coding (AVC)), the High Efficiency Video Coding (HEVC) standard currently under development, and extensions of these standards. By implementing these video compression techniques, video devices can more efficiently transmit, receive, encode, decode, and/or store digital video information.

Video compression techniques perform spatial (intra-picture) prediction and/or temporal (inter-picture) prediction to reduce or remove redundancy inherent in video sequences. For block-based video coding, a video slice (i.e., a video picture or portion of a video picture) may be partitioned into video blocks, which may also be referred to as treeblocks, coding tree blocks (CTBs), coding tree units (CTUs), coding units (CUs), and/or coding nodes. Video blocks in an intra-coded (I) slice of a picture are encoded using spatial prediction with respect to reference samples in neighboring blocks in the same picture. Video blocks in an inter-coded (P or B) slice of a picture may use spatial prediction with respect to reference samples in neighboring blocks in the same picture or temporal prediction with respect to reference samples in other reference pictures. Pictures may be referred to as frames, and reference pictures may be referred to as reference frames.

Spatial prediction or temporal prediction results in coding a predictive block for the block. Residual data represents the pixel differences between the original block to be coded and the predictive block. Inter-coded blocks are encoded based on a motion vector pointing to a block of reference samples forming the predictive block and residual data indicating the difference between the coded block and the predictive block. Intra-coded blocks are encoded based on an intra-coding mode and the residual data. For further compression, the residual data can be transformed from the pixel domain to the transform domain, thereby generating residual transform coefficients, which can then be quantized. The quantized transform coefficients, initially arranged in a two-dimensional array, can be scanned to generate a one-dimensional vector of transform coefficients, and entropy coding can be applied to achieve even more compression.

Summary of the Invention

In general, this disclosure describes techniques related to deriving a reference picture set for use in video coding. For example, the reference picture set may constitute a combination of multiple reference picture subsets. Each of the reference picture subsets may identify multiple potential reference pictures, but fewer than all potential reference pictures. In the example techniques described in this disclosure, a video coder (encoder or decoder) may construct multiple lists, each list including identifiers of a subset of the potential reference pictures. From these multiple lists, the video coder may construct the multiple reference picture subsets, which results in the video coder deriving the reference picture set.

In addition to techniques related to deriving the reference picture set, this disclosure also describes simplified reference picture list initialization techniques. This reference picture list initialization can eliminate the need to reorder the reference pictures. For example, if reference picture list modification is not required, the initial reference picture list can form the final reference picture list, and no further reordering may be required. The techniques may also relate to constructing the reference picture list in a manner in which the video coder repeatedly adds reference pictures to the reference picture list until the number of entries in the reference picture list equals the maximum allowable number of entries.

In some examples, the techniques relate to reference picture list modification. For example, the video coder may modify the initial reference picture list by referencing one or more pictures in the reference picture subset and including one or more pictures in the reference picture subset in the reference picture list after constructing the initial reference picture list.

In some examples, the video coder may perform decoded picture buffer (DPB) management. In these examples, if a decoded picture does not belong to the reference picture set, the video coder may remove the decoded picture from the DPB. In some examples, the video coder may remove the decoded picture before decoding the current picture.

In one example, the disclosure describes a method for coding video data, the method comprising coding information indicating reference pictures belonging to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The method further comprises: constructing a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures of the reference picture set; and coding the current picture based on the plurality of reference picture subsets.

In one example, the disclosure describes a device for decoding video data. The device includes a video coder configured to decode information indicating reference images belonging to a reference picture set. In this example, the reference picture set identifies the reference images that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The video coder is further configured to: construct a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures of the reference picture set; and decode the current picture based on the plurality of reference picture subsets.

In one example, the present disclosure describes a computer-readable storage medium having instructions stored thereon that, when executed, cause a processor of a device for decoding video data to decode information indicating reference images belonging to a reference picture set. In this example, the reference picture set identifies the reference images that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The instructions further cause the processor to: construct a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures of the reference picture set; and decode the current picture based on the plurality of reference picture subsets.

In one example, the disclosure describes a device for coding video data. The device includes means for coding information indicating reference pictures belonging to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The device also includes means for constructing a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures of the reference picture set; and means for coding the current picture based on the plurality of reference picture subsets.

In one example, the disclosure describes a method for decoding video data, the method comprising decoding information indicating reference pictures belonging to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The method further comprises: constructing a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures of the reference picture set; adding a reference picture from a first subset of the plurality of reference picture subsets followed by a reference picture from a second subset of the plurality of reference picture subsets, and then followed by a reference picture from a third subset of the plurality of reference picture subsets, as long as the number of reference picture list entries is not greater than a maximum number of allowable reference picture list entries; and decoding the current picture based on the reference picture list.

In one example, the disclosure describes a device for decoding video data. The device includes a video decoder configured to decode information indicating reference pictures belonging to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The video decoder is further configured to: construct a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures of the reference picture set; add the following reference pictures to a reference picture list: a reference picture from a first subset of the plurality of reference picture subsets followed by a reference picture from a second subset of the plurality of reference picture subsets, and then followed by a reference picture from a third subset of the plurality of reference picture subsets, as long as the number of reference picture list entries is not greater than a maximum number of allowable reference picture list entries; and decode the current picture based on the reference picture list.

In one example, the present disclosure describes a computer-readable storage medium having instructions stored thereon that, when executed, cause a processor of a device for decoding video data to decode information indicating reference pictures belonging to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The instructions further cause the processor to: construct a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures of the reference picture set; add each of the following reference pictures to a reference picture list: a reference picture from a first subset of the plurality of reference picture subsets followed by a reference picture from a second subset of the plurality of reference picture subsets, and then followed by a reference picture from a third subset of the plurality of reference picture subsets, as long as the number of reference picture list entries is not greater than a maximum number of allowable reference picture list entries; and decode the current picture based on the reference picture list.

In one example, the disclosure describes a device for decoding video data. The device includes means for decoding information indicating reference pictures belonging to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The device also includes: means for constructing a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures of the reference picture set; means for adding to a reference picture list a reference picture from a first subset of the plurality of reference picture subsets followed by a reference picture from a second subset of the plurality of reference picture subsets, and followed by a reference picture from a third subset of the plurality of reference picture subsets, as long as the number of reference picture list entries is not greater than a maximum number of allowable reference picture list entries; and means for decoding the current picture based on the reference picture list.

In one example, this disclosure describes a method for coding video data that includes coding information indicating a reference picture that belongs to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The method also includes: constructing multiple reference image subsets, each reference image subset identifying zero or more of the reference images of the reference image set; adding reference images from the multiple reference image subsets to a first set of entries in a reference image list; determining whether the number of entries in the reference image list is equal to the maximum number of allowable entries in the reference image list; when the number of entries in the reference image list is not equal to the maximum number of allowable entries in the reference image list, repeatedly re-adding one or more reference images from at least one of the reference image subsets to the entries in the reference image list after the first set of entries until the number of entries in the reference image list is equal to the maximum number of allowable entries in the reference image list; and decoding the current image based on the reference image list.

In one example, the disclosure describes a device for coding video data. The device includes a video coder configured to decode information indicating reference pictures belonging to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The video decoder is also configured to: construct multiple reference image subsets, each reference image subset identifying zero or more of the reference images of the reference image set; add reference images from the multiple reference image subsets to a first set of entries in a reference image list; determine whether the number of entries in the reference image list is equal to the maximum number of allowable entries in the reference image list; when the number of entries in the reference image list is not equal to the maximum number of allowable entries in the reference image list, repeatedly re-add one or more reference images from at least one of the reference image subsets to the entries in the reference image list that follow the first set of entries until the number of entries in the reference image list is equal to the maximum number of allowable entries in the reference image list; and decode the current image based on the reference image list.

In one example, this disclosure describes a computer-readable storage medium having stored thereon instructions that, when executed, cause a processor of a device for decoding video data to decode information indicating reference pictures belonging to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-predicting a current picture and can potentially be used for inter-predicting one or more pictures that follow the current picture in decoding order. The instructions also cause the processor to: construct multiple reference image subsets, each reference image subset identifying zero or more of the reference images of the reference image set; add reference images from the multiple reference image subsets to a first set of entries in a reference image list; determine whether the number of entries in the reference image list is equal to the maximum number of allowable entries in the reference image list; when the number of entries in the reference image list is not equal to the maximum number of allowable entries in the reference image list, repeatedly re-add one or more reference images from at least one of the reference image subsets to the entries in the reference image list that follow the first set of entries until the number of entries in the reference image list is equal to the maximum number of allowable entries in the reference image list; and decode the current image based on the reference image list.

In one example, the disclosure describes a device for coding video data. The device includes means for coding information indicating reference pictures belonging to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-predicting a current picture and can potentially be used for inter-predicting one or more pictures that follow the current picture in decoding order. The device also includes: a device for constructing multiple reference image subsets, each reference image subset identifying zero or more of the reference images of the reference image set; a device for adding reference images from the multiple reference image subsets to a first set of entries in a reference image list; a device for determining whether the number of entries in the reference image list is equal to the maximum number of allowable entries in the reference image list; when the number of entries in the reference image list is not equal to the maximum number of allowable entries in the reference image list, a device for repeatedly re-adding one or more reference images from at least one of the reference image subsets to the entries in the reference image list that follow the first set of entries until the number of entries in the reference image list is equal to the maximum number of allowable entries in the reference image list; and a device for decoding the current image based on the reference image list.

In one example, the disclosure describes a method for decoding video data, the method comprising decoding information indicating reference pictures belonging to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The method further comprises: constructing a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures of the reference picture set; constructing an initial reference picture list based on the constructed reference picture subsets; and when reference picture modification is required, identifying a reference picture in at least one of the constructed reference picture subsets; and adding the identified reference picture to a current entry in the initial reference picture list to construct a modified reference picture list. The method further comprises decoding the current picture based on the modified reference picture list.

In one example, the disclosure describes a device for decoding video data. The device includes a video decoder configured to decode information indicating reference images belonging to a reference picture set. In this example, the reference picture set identifies the reference images that can potentially be used for inter-frame prediction of a current picture and can potentially be used for inter-frame prediction of one or more pictures that follow the current picture in decoding order. The video decoder is further configured to: construct a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures of the reference picture set; construct an initial reference picture list based on the constructed reference picture subsets; and when reference picture modification is required, identify a reference picture in at least one of the constructed reference picture subsets; and add the identified reference picture to a current entry in the initial reference picture list to construct a modified reference picture list. The video decoder is also configured to decode the current picture based on the modified reference picture list.

In one example, the present disclosure describes a computer-readable storage medium having instructions stored thereon that, when executed, cause a processor of a device for decoding video data to decode information indicating reference images belonging to a reference picture set. In this example, the reference picture set identifies the reference images that can potentially be used for inter-frame prediction of a current picture and can potentially be used for inter-frame prediction of one or more pictures that follow the current picture in decoding order. The instructions further cause the processor to: construct a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures of the reference picture set; construct an initial reference picture list based on the constructed reference picture subsets; and, when reference picture modification is required, identify a reference picture in at least one of the constructed reference picture subsets; and add the identified reference picture to a current entry in the initial reference picture list to construct a modified reference picture list. The instructions further cause the processor to decode the current picture based on the modified reference picture list.

In one example, the disclosure describes a device for decoding video data. The device includes means for decoding information indicating reference images belonging to a reference image set. In this example, the reference image set identifies the reference images that can potentially be used for inter-frame prediction of a current image and can potentially be used for inter-frame prediction of one or more images that follow the current image in decoding order. The device also includes: means for constructing a plurality of reference image subsets, each reference image subset identifying zero or more of the reference images of the reference image set; means for constructing an initial reference image list based on the constructed reference image subsets; and means for identifying a reference image in at least one of the constructed reference image subsets when reference image modification is required; and means for adding the identified reference image to a current entry in the initial reference image list to construct a modified reference image list. The device also includes means for decoding the current image based on the modified reference image list.

In one example, the disclosure describes a method for coding video data, the method comprising coding information indicating reference pictures belonging to a reference picture set. In this example, the reference picture set identifies the reference pictures that can potentially be used for inter-prediction of a current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The method comprises: deriving the reference picture set based on the coded information; determining whether a decoded picture stored in a decoded picture buffer (DPB) is not required for output and is not identified in the reference picture set; when the decoded picture is not required for output and is not identified in the reference picture set, removing the decoded picture from the DPB; and, after removing the decoded picture, decoding the current picture.

In one example, the disclosure describes a device for decoding video data. The device includes a video decoder configured to decode information indicating reference images belonging to a reference picture set. In this example, the reference picture set identifies the reference images that can potentially be used for inter-frame prediction of a current picture and can potentially be used for inter-frame prediction of one or more pictures that follow the current picture in decoding order. The video decoder is further configured to: derive the reference picture set based on the decoded information; determine whether a decoded picture stored in a decoded picture buffer (DPB) is not required for output and is not identified in the reference picture set; when the decoded picture is not required for output and is not identified in the reference picture set, remove the decoded picture from the DPB; and, after removing the decoded picture, decode the current picture.

In one example, the present disclosure describes a computer-readable storage medium having instructions stored thereon that, when executed, cause a processor of a device for decoding video data to decode information indicating reference images belonging to a reference picture set. In this example, the reference picture set identifies reference images that can potentially be used for inter-frame prediction of a current picture and can potentially be used for inter-frame prediction of one or more pictures that follow the current picture in decoding order. The instructions further cause the processor to: derive the reference picture set based on the decoded information; determine whether a decoded picture stored in a decoded picture buffer (DPB) is not required for output and is not identified in the reference picture set; if the decoded picture is not required for output and is not identified in the reference picture set, remove the decoded picture from the DPB; and, after removing the decoded picture, decode the current picture.

In one example, the disclosure describes a device for decoding video data. The device includes means for decoding information indicating reference images belonging to a reference picture set. In this example, the reference picture set identifies the reference images that can potentially be used for inter-frame prediction of a current picture and can potentially be used for inter-frame prediction of one or more pictures that follow the current picture in decoding order. The device also includes: means for deriving the reference picture set based on the decoded information; means for determining whether a decoded picture stored in a decoded picture buffer (DPB) does not need to be output and is not identified in the reference picture set; means for removing the decoded picture from the DPB when the decoded picture does not need to be output and is not identified in the reference picture set; and means for decoding the current picture after removing the decoded picture.

In one example, the disclosure describes a method of coding video data, the method comprising coding a syntax element indicating candidate long-term reference pictures identified in a parameter set. In this example, one or more of the candidate long-term reference pictures belong to a reference picture set for a current picture. Moreover, in this example, the reference picture set identifies reference pictures that can potentially be used for inter-prediction of the current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The method further comprises: coding a syntax element indicating which candidate long-term reference pictures identified in the parameter set belong to the reference picture set for the current picture; and constructing at least one of a plurality of reference picture subsets based on the indication of which candidate long-term reference pictures belong to the reference picture set for the current picture. In this example, the plurality of reference picture subsets form the reference picture set.

In one example, the disclosure describes a device for decoding video data. The device includes a video decoder configured to decode a syntax element indicating candidate long-term reference pictures identified in a parameter set. In this example, one or more of the candidate long-term reference pictures belong to a reference picture set of a current picture. Moreover, in this example, the reference picture set identifies reference pictures that can potentially be used for inter-frame prediction of the current picture and can potentially be used for inter-frame prediction of one or more pictures that follow the current picture in decoding order. The video decoder is further configured to: decode a syntax element indicating which candidate long-term reference pictures identified in the parameter set belong to the reference picture set of the current picture; and construct at least one of a plurality of reference picture subsets based on the indication of which candidate long-term reference pictures belong to the reference picture set of the current picture. In this example, the plurality of reference picture subsets form the reference picture set.

In one example, the disclosure describes a computer-readable storage medium having instructions stored thereon that, when executed, cause a processor of a device for decoding video data to decode a syntax element indicating candidate long-term reference pictures identified in a parameter set. In this example, one or more of the candidate long-term reference pictures belong to a reference picture set for a current picture. Furthermore, in this example, the reference picture set identifies reference pictures that can potentially be used for inter-prediction of the current picture and can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order. The instructions further cause the processor to: decode a syntax element indicating which candidate long-term reference pictures identified in the parameter set belong to the reference picture set for the current picture; and construct at least one of a plurality of reference picture subsets based on the indication of which candidate long-term reference pictures belong to the reference picture set for the current picture. In this example, the plurality of reference picture subsets form the reference picture set.

In one example, the disclosure describes a device for decoding video data. The device includes means for decoding a syntax element indicating candidate long-term reference pictures identified in a parameter set. In this example, one or more of the candidate long-term reference pictures belong to a reference picture set of a current picture. Moreover, in this example, the reference picture set identifies reference pictures that can potentially be used for inter-frame prediction of the current picture and can potentially be used for inter-frame prediction of one or more pictures that follow the current picture in decoding order. The device also includes: means for decoding a syntax element indicating which candidate long-term reference pictures identified in the parameter set belong to the reference picture set of the current picture; and means for constructing at least one of a plurality of reference picture subsets based on the indication of which candidate long-term reference pictures belong to the reference picture set of the current picture. In this example, the plurality of reference picture subsets form the reference picture set.

The details of one or more examples are set forth in the accompanying drawings and the description below. Other features, objects, and advantages will be apparent from the description and drawings, and from the claims.

BRIEF DESCRIPTION OF THE DRAWINGS

1 is a block diagram illustrating an example video encoding and decoding system that may utilize the techniques described in this disclosure.

2 is a conceptual diagram illustrating an example video sequence including multiple pictures that are encoded and transmitted.

3 is a block diagram illustrating an example video encoder that may implement the techniques described in this disclosure.

4 is a block diagram illustrating an example video decoder that may implement the techniques described in this disclosure.

5 is a flow diagram illustrating an example operation of deriving a reference image set.

6 is a flow diagram illustrating an example operation of constructing a reference picture list.

7 is a flow diagram illustrating another example operation of constructing a reference picture list.

8 is a flow diagram illustrating an example operation of modifying an initial reference picture list.

9 is a flow diagram illustrating an example operation of decoded image removal.

10 is a flow diagram illustrating an example operation of determining which long-term reference pictures belong to a reference picture set for a current picture.

DETAILED DESCRIPTION

The techniques of this disclosure generally relate to the management of reference pictures used for inter-frame prediction. For example, a video coder (e.g., a video encoder or decoder) includes a decoded picture buffer (DPB). The DPB stores decoded pictures, including reference pictures. Reference pictures are pictures that can potentially be used for inter-frame prediction of a picture. In other words, during the decoding (encoding or decoding) of a picture, the video coder can predict the picture based on one or more reference pictures stored in the DPB.

In order to efficiently utilize the DPB, a DPB management process may be specified, such as a process for storing decoded pictures in the DPB, a process for marking reference pictures, a process for outputting and removing decoded pictures from the DPB, etc. Generally speaking, in some current and developing video coding standards, DPB management may include one or more of the following aspects: picture identification and reference picture identification, reference picture list construction, reference picture marking, picture output from the DPB, picture insertion into the DPB, and picture removal from the DPB.

To aid understanding, a brief description of how reference picture marking and reference picture list construction may occur according to some video coding standards is provided below. Some of the techniques described in this disclosure address issues that may exist in reference picture marking, reference picture list construction, and DPB picture removal and output in order to improve the efficiency of utilizing the DPB.

For reference picture marking, the maximum number of reference pictures used for inter-frame prediction (called M (num_ref_frames)) is indicated in the active sequence parameter set. When a reference picture is decoded, it is marked as "used for reference". If the decoding of the reference picture results in more than M pictures being marked as "used for reference", then at least one picture must be marked as "not used for reference". If the pictures marked as "not used for reference" are not yet needed for output, the DPB removal process removes the pictures marked as "not used for reference" from the DPB.

When a picture is decoded, it may be a non-reference picture or a reference picture. A reference picture may be a long-term reference picture or a short-term reference picture, and when a reference picture is marked as "unused for reference", it may no longer be needed for reference. In some video coding standards, there may be a reference picture marking operation that changes the status of a reference picture.

Two types of operations are possible for reference picture marking: sliding window and adaptive memory control. The operating mode for reference picture marking can be selected on a picture-by-picture basis; however, the sliding window operation operates in a first-in, first-out queue with a fixed number of short-term reference pictures. In other words, the short-term reference picture with the oldest decoding time (the picture marked as unused for reference) can be implicitly removed first.

However, adaptive memory control explicitly removes short-term or long-term pictures. Adaptive memory control also enables switching the status of short-term and long-term pictures, among other things. For example, in adaptive memory control, a video encoder may signal a syntax element that specifies which pictures should be marked for reference. A video decoder may receive the syntax element and mark the pictures as specified. In a sliding window, the video encoder may not need to signal which pictures should be marked for reference. Instead, the video decoder may implicitly (i.e., without receiving the syntax element) determine which pictures should be marked for reference based on which pictures are within the sliding window.

The video coder may also be tasked with constructing reference picture lists that indicate which reference pictures may be used for inter-frame prediction purposes. Two of these reference picture lists are referred to as List 0 and List 1, respectively. The video coder first constructs List 0 and List 1 using a default construction technique (e.g., a pre-configured construction scheme for constructing List 0 and List 1). Optionally, after constructing the initial List 0 and List 1, the video decoder may decode syntax elements that, when present, may direct the video decoder to modify the initial List 0 and List 1.

The video encoder may signal a syntax element indicating an identifier of a reference picture in the DPB. The video encoder may also signal a syntax element containing an index within List 0, List 1, or both List 0 and List 1, indicating which reference picture or pictures are to be used to decode a coded block of a current picture. In turn, the video decoder uses the received identifier to identify one or more index values of one or more reference pictures listed in List 0, List 1, or both List 0 and List 1. Based on the index value(s) and identifier(s) of the one or more reference pictures, the video decoder retrieves the one or more reference pictures, or portions of the one or more reference pictures, from the DPB and decodes a coded block of the current picture based on the one or more retrieved reference pictures and one or more motion vectors identifying blocks within the one or more reference pictures used to decode the coded block.

For example, reference picture list construction for the first or second reference picture lists of bidirectionally predicted pictures involves two steps: reference picture list initialization and reference picture list modification (also referred to as reference picture list reordering). Reference picture list initialization can be an implicit mechanism that places reference pictures in a reference picture memory (also referred to as a decoded picture buffer) into a list based on their Picture Order Count (POC) values (aligned with the display order of the pictures). The reference picture list reordering mechanism can modify the positions of pictures placed in the list during reference picture list initialization to any new position, or even place any reference picture in the reference picture memory at any position if the picture did not belong to the initialized list. After reference picture list reordering (modification), some pictures may be placed very far into the list. However, if a picture's position exceeds the number of valid reference pictures in the list, it will not be considered an entry in the final reference picture list. The number of valid reference pictures for a list can be signaled in the slice header of each list.

The techniques described in this disclosure may be applicable to various video coding standards. Examples of video coding standards include ITU-T H.261, ISO/IEC MPEG-1 Video, ITU-T H.262 or ISO/IEC MPEG-2 Video, ITU-T H.263, ISO/IEC MPEG-4 Video, and ITU-T H.264 (also known as ISO/IEC MPEG-4 AVC), including its Scalable Video Coding (SVC) and Multi-view Video Coding (MVC) extensions. Additionally, there is a new video coding standard, High Efficiency Video Coding (HEVC), being developed by the Joint Collaboration Team on Video Coding (JCT-VC) of the ITU-T Video Coding Experts Group (VCEG) and the ISO/IEC Motion Picture Experts Group (MPEG).

For illustrative purposes only, the techniques are described in the context of the HEVC standard. As of 20 July 2012, the latest working draft (WD) of HEVC is available from http://phenix.int-evry.fr/jct/doc_end_user/documents/10_Stockholm/wg11/JCTVC-J1003-v8.zip (and referred to hereinafter as HEVC WD8).

As described above, the techniques described in this disclosure may address problems that may exist in existing solutions for decoded picture buffer (DPB) management. As an example, in some example techniques described in this disclosure, marking of reference pictures as "not used for reference" may not be required. For example, the techniques described in this disclosure may address the following issues: issues related to DPB management techniques that are not well suited for temporal scalability, issues related to signaling overhead for long-term reference pictures, and issues related to efficiency and complexity of reference picture list initialization and modification. The techniques described in this disclosure may also address the following issues: issues related to marking of "no reference picture" for non-completed entries in a reference picture list during reference picture list initialization; issues related to decoded picture output, insertion into the DPB, and removal from the DPB; and issues related to possible values for picture sequence number (POC) values.

According to the techniques described in this disclosure, a reference picture list is constructed from a reference picture set. A reference picture set is defined as a set of reference pictures associated with a picture, consisting of all reference pictures that precede the associated picture in decoding order, which can be used for inter-prediction of blocks in the associated picture or for any pictures that follow the associated picture in decoding order, e.g., until the next instantaneous decoding refresh (IDR) picture or broken link access (BLA) picture. In other words, the reference pictures in the reference picture set may require the following properties: (1) they all precede the current picture in decoding order, and (2) they can be used for inter-prediction of the current picture and/or for inter-prediction of any pictures that follow the current picture in decoding order, and in some examples, until the next IDR picture or BLA picture. Other alternative definitions of reference picture sets may exist, which are provided below.

In the example techniques described in this disclosure, a video coder may derive a reference picture set, and after such derivation, the video coder may construct a reference picture list. For example, only reference pictures in the reference picture set may be candidate reference pictures for constructing the reference picture list.

To construct a reference picture set, a video coder may construct multiple reference picture subsets. A combination of reference picture subsets may together form a reference picture set. For example, a video encoder may explicitly signal in the coded bitstream a value for an identifier that allows a video decoder to determine the reference pictures included in the reference picture set. For example, the identifier of a reference picture may be a picture sequence number. Each picture is associated with a picture sequence number (referred to as PicOrderCnt). PicOrderCnt indicates the output order or display order of the corresponding picture relative to the previous IDR picture in decoding order, and, in some other alternatives, indicates the output order position of the associated picture relative to the output order position of other pictures in the same coded video sequence.

PicOrderCnt may be referred to as a picture order code (POC) value. The POC value may indicate the output or display order of pictures and may be used to identify pictures. For example, within a coded video sequence, pictures with smaller POC values are output or displayed earlier than pictures with larger POC values.

The video decoder may determine identifiers for the reference images and construct the plurality of reference image subsets from these identifiers. Based on these reference image subsets, the video decoder may derive reference image sets, as described in more detail below. In some examples, each of the reference image subsets includes a different reference image because the reference images in the reference image subsets do not overlap. In this manner, each of the reference images may be present in only one of the reference image subsets and not in any other reference image subsets. However, aspects of this disclosure should not be considered so limited.

After determining identifiers (e.g., POC values) for reference pictures in a reference picture set or a subset of a reference picture set, the video decoder may construct reference picture subsets. As described in more detail below, the video decoder may construct six reference picture subsets, but the video decoder may be able to construct more or fewer reference picture subsets.

These six reference picture subsets are named: RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, RefPicSetStFoll1, RefPicSetLtCurr, and RefPicSetLtFoll. The RefPicSetStCurr0 reference picture subset may be referred to as the RefPicSetStCurrBefore reference picture subset, and the RefPicSetStCurr1 reference picture subset may be referred to as the RefPicSetStCurrAfter reference picture subset.

The RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, and RefPicSetStFoll1 reference picture subsets may identify short-term reference pictures. In some examples, these reference picture subsets may identify short-term reference pictures based on whether the short-term reference picture is earlier in display order than the current picture being coded or later in display order than the current picture being coded, and whether the short-term reference picture can potentially be used for inter-prediction of the current picture and pictures that follow the current picture in decoding order, or can potentially be used for inter-prediction of only pictures that follow the current picture in decoding order.

For example, the RefPicSetStCurr0 reference picture subset may include (and may only include) identification information (e.g., POC values) of all short-term reference pictures that have an output or display order earlier than that of the current picture and that are potentially usable for reference in inter-frame prediction of the current picture and that are potentially usable for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order. The RefPicSetStCurr1 reference picture subset may include (and may only include) identification information of all short-term reference pictures that have an output or display order later than that of the current picture and that are potentially usable for reference in inter-frame prediction of the current picture and that are potentially usable for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order.

The RefPicSetStFoll0 reference picture subset may include (and may only include) identification information of all short-term reference pictures that satisfy the following conditions: have an output or display order earlier than that of the current picture, may be potentially used for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order, and cannot be used for reference in inter-frame prediction of the current picture. The RefPicSetStFoll1 reference picture subset may include (and may only include) identification information of all short-term reference pictures that satisfy the following conditions: have an output or display order later than that of the current picture, may be potentially used for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order, and cannot be used for reference in inter-frame prediction of the current picture.

The RefPicSetLtCurr and RefPicSetLtFoll reference picture subsets may identify long-term reference pictures. In some examples, these reference picture subsets may identify long-term reference pictures based on whether the long-term reference picture is earlier in display order than the current picture being coded or later in display order than the current picture being coded.

For example, the RefPicSetLtCurr reference picture subset may include (and may only include) identification information of all long-term reference pictures that can potentially be used for reference in inter-frame prediction of the current picture and can potentially be used for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order. The RefPicSetLtFoll reference picture subset may include (and may only include) identification information of all long-term reference pictures that can potentially be used for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order and cannot be used for reference in inter-frame prediction of the current picture.

After constructing the reference picture subsets, a video decoder may order the reference picture subsets in different orders to derive a reference picture set. As an example, the order of the reference picture sets may be RefPicSetStCurr0, RefPicSetSetCurr1, RefPicSetFoll0, RefPicSetFoll1, RefPicSetLtCurr, and RefPicSetLtFoll. However, other orderings of the subsets may be possible to derive a reference picture set. For instance, as another example, the order of the reference picture sets may be RefPicSetStCurr0 reference picture subset, followed by RefPicSetStCurr1 reference picture subset, followed by RefPicSetLtCurr reference picture subset, followed by RefPicSetStFoll0 reference picture subset, followed by RefPicSetFoll1 reference picture subset, and followed by RefPicSetLtFoll reference picture subset.

According to the techniques described in this disclosure, the RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr subsets include all reference pictures that can be used in inter-prediction of blocks in the current picture and that can be used in inter-prediction of one or more of the pictures that follow the current picture in decoding order. The RefPicSetStFoll0, RefPicSetStFoll1, and RefPicSetLtFoll subsets include all reference pictures that are not used in inter-prediction of blocks in the current picture and that can be used in inter-prediction of one or more of the pictures that follow the current picture in decoding order.

It should be understood that six reference image subsets are described for illustrative purposes only and should not be considered limiting. In alternative examples, there may be more or fewer reference image subsets. These reference image subsets in these alternative examples are described in more detail below.

In some techniques described in these disclosures, a video decoder may not need to mark decoded pictures as "used for reference," "not used for reference," "used for short-term reference," or "used for long-term reference." Instead, whether a decoded picture stored in the DPB requires inter-frame prediction is indicated by whether it is included in the reference picture set of the current picture. In alternative examples, the video decoder may have the possibility to mark decoded pictures as "used for reference," "not used for reference," "used for short-term reference," or "used for long-term reference." In these examples, after the video decoder decodes a picture, the decoded picture is a reference picture and is marked as "used for reference." Then, after invoking the process for reference picture set derivation, and before possible removal of the decoded picture from the DPB, all reference pictures stored in the DPB but not included in the reference picture set of the current picture are marked as "not used for reference." Thus, whether a decoded picture stored in the DPB requires inter-frame prediction can be indicated by whether it is marked as "used for reference."

Once the video decoder derives a reference picture set from the plurality of reference picture subsets, the video decoder may construct reference picture lists (e.g., List 0 and List 1) from the reference picture set. For example, construction of the reference picture lists may include an initialization step and possibly a modification step. By deriving the reference picture set in the manner described above, the video decoder may be able to improve the efficiency and reduce the complexity of reference picture list initialization and reference picture list modification.

There are various ways in which a video decoder can construct a reference picture list. The techniques described in this disclosure provide a mechanism by which a video decoder can construct a reference picture list without reordering the reference pictures to be included in the (initial) reference picture list. For example, a video decoder can be configured to implement a default reference list construction technique, in which the video decoder constructs an initial reference picture list using a subset of reference pictures. Then, if no reference picture list modification is required, the final reference picture list can be identical to the initial reference picture list, without requiring any additional reordering of the reference picture list.

In some examples, the techniques described in this disclosure may relate to constructing a reference picture list in a manner that eliminates incomplete entries. For example, the techniques may repeatedly add one or more reference pictures from a reference picture subset to a reference picture list. For example, after a video decoder adds one or more reference pictures from a reference picture subset used to construct an initial reference picture list, the video decoder may determine whether the number of entries in the reference picture list is less than a maximum allowable number of entries. If the number of entries in the reference picture list is less than the maximum allowable number of entries, the video decoder may re-add at least one of the reference pictures from one of the reference picture subsets used to construct the reference picture list to the reference picture list. This re-addition of the reference picture (also referred to as re-listing) may occur at a different location within the reference picture list than the location where the video decoder first added the reference picture.

As used herein, "relisting" or "re-adding" refers to re-adding (e.g., re-identifying) a previously added (e.g., identified) reference image to the initial reference image list. However, when a reference image is re-added, the reference image may be located at two different entries in the initial reference image list. In other words, when a reference image is re-added, there may be two index values in the initial reference image list that identify the same reference image.

In some examples, the techniques described in this disclosure may relate to modifying an initial reference picture list. For example, a video decoder may construct an initial reference picture list. The video decoder may determine that a reference picture list modification is required based on syntax elements signaled by the video encoder in the coded bitstream. When a reference picture list modification is required, the video decoder may identify a reference picture in at least one of the constructed reference picture subsets. The video decoder may list (e.g., add) the identified reference picture in a current entry of the initial reference picture list to construct a modified reference picture list. The video decoder may then decode a current picture based on the modified reference picture list.

In some examples, the techniques described in this disclosure may relate to outputting and removing decoded pictures from a decoded picture buffer (DPB). The example techniques may remove a decoded picture from the DPB before coding a current picture. For example, if the decoded picture is not identified in the reference picture set of the current picture and if the decoded picture does not need to be output (i.e., the decoded picture is not intended to be output or the decoded picture is intended to be output but has already been output), the example techniques may remove the decoded picture.

FIG1 is a block diagram illustrating an example video encoding and decoding system 10 that may utilize the techniques described in this disclosure. Generally speaking, a reference picture set is defined as a set of reference pictures associated with a picture that consists of all reference pictures that precede the associated picture in decoding order and that may be used for inter-prediction of the associated picture or any pictures that follow the associated picture in decoding order. In some examples, the reference pictures that precede the associated picture may be reference pictures up to the next instantaneous decoded refresh (IDR) picture or broken link attached (BLA) picture. In other words, the reference pictures in the reference picture set may all precede the current picture in decoding order. Furthermore, the reference pictures in the reference picture set may be used for inter-prediction of the current picture and/or for inter-prediction of any pictures that follow the current picture in decoding order, up to the next IDR or BLA picture.

Other alternative definitions of a reference picture set are possible. For example, a reference picture set may be a set of reference pictures associated with a picture, consisting of all reference pictures (excluding the associated picture itself) that can be used for inter-frame prediction of the associated picture or any picture following the associated picture in decoding order, and that have a temporal_id that is less than or equal to the temporal_id of the associated picture. The temporal_id may be a temporal identification value. The temporal identification value may be a hierarchical value that indicates which pictures can be used to code the current picture. In general, a picture with a particular temporal_id value may potentially be a reference picture for a picture with an equal or greater temporal_id value, but the reverse is not true. For example, a picture with a temporal_id value of 1 may potentially be a reference picture for pictures with temporal_id values of 1, 2, 3, ..., but not for a picture with a temporal_id value of 0.

The lowest temporal_id value may also indicate the lowest display rate. For example, if the video decoder only decodes pictures with a temporal_id value of 0, the display rate may be 7.5 pictures/second. If the video decoder only decodes pictures with temporal_id values 0 and 1, the display rate may be 15 pictures/second, and so on.

As another example, a reference picture set may be a reference picture set associated with a picture, consisting of all reference pictures (excluding the associated picture itself) that can be used for inter-prediction of the associated picture or any picture that follows the associated picture in decoding order. As yet another example, a reference picture set may be defined as a reference picture set associated with a picture, consisting of all reference pictures (possibly including the associated picture itself) that can be used for inter-prediction of the associated picture or any picture that follows the associated picture in decoding order. As another example, a reference picture set may be defined as a reference picture set associated with a picture, consisting of all reference pictures (possibly including the associated picture itself) that can be used for inter-prediction of the associated picture or any picture that follows the associated picture in decoding order, and having a temporal_id that is less than or equal to the temporal_id of the associated picture.

As yet another example, in the above definition of a reference picture set, the phrase “usable for inter prediction” is replaced with “used for inter prediction.” Although alternative definitions of a reference picture set are possible, in this disclosure, an example is described with the following definition of a reference picture set: a reference picture set is a set of reference pictures associated with a picture, consisting of all reference pictures that precede the associated picture in decoding order, and that can be used for inter prediction of the associated picture or any picture that follows the associated picture in decoding order.

For example, some of the reference images in the reference image set are reference images that can potentially be used to inter-predict blocks of the current image and cannot be used to inter-predict images that follow the current image in decoding order. Some of the reference images in the reference image set are reference images that can potentially be used to inter-predict blocks of the current image and blocks in one or more images that follow the current image in decoding order. Some of the reference images in the reference image set are reference images that can potentially be used to inter-predict blocks in one or more images that follow the current image in decoding order and cannot be used to inter-predict blocks in the current image.

As used in this disclosure, a reference picture potentially usable for inter-frame prediction refers to a reference picture that can be used for inter-frame prediction, but is not necessarily required for inter-frame prediction. For example, a reference picture set may identify reference pictures potentially usable for inter-frame prediction. However, this does not mean that all identified reference pictures are necessarily required for inter-frame prediction. Rather, one or more of these identified reference pictures may be usable for inter-frame prediction, but not all of them are necessarily required for inter-frame prediction.

1 , system 10 includes a source device 12 that generates encoded video for decoding by a destination device 14. Source device 12 and destination device 14 may each be an example of a video coding device. Source device 12 may transmit the encoded video to destination device 14 via a communication channel 16 or may store the encoded video on a storage medium 17 or a file server 19 so that the encoded video can be accessed by destination device 14 as needed.

Source device 12 and destination device 14 may comprise any of a wide range of devices, including wireless phones such as so-called "smart" phones, so-called "smart" pads, or other such wireless devices equipped for wireless communication. Additional examples of source device 12 and destination device 14 include, but are not limited to, digital televisions, devices in digital live broadcast systems, devices in wireless broadcast systems, personal digital assistants (PDAs), laptop computers, desktop computers, tablet computers, e-book readers, digital cameras, digital recording devices, digital media players, video game devices, video game consoles, cellular radiotelephones, satellite radiotelephones, video teleconferencing devices, and video streaming devices, wireless communication devices, or the like.

As indicated above, in many cases, source device 12 and/or destination device 14 may be equipped for wireless communication. Thus, communication channel 16 may comprise a wireless channel, a wired channel, or a combination of wireless and wired channels suitable for transmitting encoded video data. Similarly, file server 19 may be accessed by destination device 14 through any standard data connection, including an Internet connection. This data connection may include a wireless channel (e.g., a Wi-Fi connection), a wired connection (e.g., DSL, cable modem, etc.) suitable for accessing encoded video data stored on the file server, or a combination of both wireless and wired channels.

However, the techniques of this disclosure may be applicable to video coding to support any of a variety of multimedia applications, such as over-the-air television broadcasting, cable television transmission, satellite television transmission, streaming video transmission (e.g., via the Internet), encoding digital video for storage on a data storage medium, decoding digital video stored on a data storage medium, or other applications. In some examples, system 10 may be configured to support one-way or two-way video transmission to support applications such as video streaming, video playback, video broadcasting, and/or video telephony.

1 , source device 12 includes a video source 18, a video encoder 20, a modulator/demodulator (modem) 22, and an output interface 24. In source device 12, video source 18 may include a source such as a video capture device (e.g., a video camera), a video archive containing previously captured video, a video feed interface that receives video from a video content provider, and/or a computer graphics system for generating computer graphics data as source video, or a combination of such sources. As one example, if video source 18 is a video camera, source device 12 and destination device 14 may form so-called camera phones or video phones. However, the techniques described in this disclosure may be applicable to video coding in general and may be applied to wireless and/or wired applications.

The captured, pre-captured, or computer-generated video may be encoded by video encoder 20. The encoded video information may be modulated by modem 22 according to a communication standard (e.g., a wireless communication protocol) and transmitted to destination device 14 via output interface 24. Modem 22 may include various mixers, filters, amplifiers, or other components designed for signal modulation. Output interface 24 may include circuits designed for transmitting data, including amplifiers, filters, and one or more antennas.

The captured, pre-captured, or computer-generated video encoded by video encoder 20 may also be stored on storage medium 17 or file server 19 for later consumption. Storage medium 17 may include a Blu-ray disc, a DVD, a CD-ROM, flash memory, or any other suitable digital storage medium for storing encoded video. The encoded video stored on storage medium 17 may then be accessed by destination device 14 for decoding and playback.

File server 19 may be any type of server capable of storing encoded video and transmitting the encoded video to destination device 14. Example file servers include a network server (e.g., for a website), an FTP server, a network attached storage (NAS) device, a local disk drive, or any other type of device capable of storing and transmitting encoded video data to a destination device. The transmission of the encoded video data from file server 19 may be streaming, downloading, or a combination of both. File server 19 may be accessed by destination device 14 through any standard data connection, including an Internet connection. This standard data connection may include a wireless channel (e.g., a Wi-Fi connection), a wired connection (e.g., DSL, cable modem, Ethernet, USB, etc.) suitable for accessing the encoded video data stored on the file server, or a combination of wireless and wired channels.

1 , destination device 14 includes an input interface 26, a modem 28, a video decoder 30, and a display device 32. Input interface 26 of destination device 14 receives information via channel 16 (as one example), or from storage medium 17 or file server 19 (as alternative examples), and modem 28 demodulates the information to produce a demodulated bitstream for video decoder 30. The demodulated bitstream may include a variety of syntax information generated by video encoder 20 for use by video decoder 30 in decoding the video data. This syntax may also be included within the encoded video data stored on storage medium 17 or file server 19. As an example, the syntax may be embedded within the encoded video data, but aspects of this disclosure should not be considered limited to this requirement. Syntax information defined by video encoder 20 and also used by video decoder 30 may include syntax elements that describe characteristics and/or processing of video blocks, such as coding tree units (CTUs), coding tree blocks (CTBs), prediction units (PUs), coding units (CUs), or other units of coded video, such as video slices, video pictures, and video sequences or groups of pictures (GOPs). Each of video encoder 20 and video decoder 30 may form part of a respective encoder-decoder (CODEC) capable of encoding or decoding video data.

Display device 32 may be integrated with destination device 14 or external to destination device 14. In some examples, destination device 14 may include an integrated display device and also be configured to interface with an external display device. In other examples, destination device 14 may be a display device. In general, display device 32 displays decoded video data to a user and may comprise any of a variety of display devices, such as a liquid crystal display (LCD), a plasma display, an organic light emitting diode (OLED) display, or another type of display device.

1 , communication channel 16 may comprise any wireless or wired communication medium, such as a radio frequency (RF) spectrum or one or more physical transmission lines, or any combination of wireless and wired media. Communication channel 16 may form part of a packet-based network, such as a local area network, a wide area network, or a global network such as the Internet. Communication channel 16 generally represents any suitable communication medium, or collection of different communication media, for transmitting video data from source device 12 to destination device 14, including any suitable combination of wired or wireless media. Communication channel 16 may include routers, switches, base stations, or any other equipment that can be used to facilitate communication from source device 12 to destination device 14.

Video encoder 20 and video decoder 30 may operate according to video compression standards, such as ITU-T H.261, ISO/IEC MPEG-1 Video, ITU-T H.262 or ISO/IEC MPEG-2 Video, ITU-T H.263, ISO/IEC MPEG-4 Video, and ITU-T H.264 (also known as ISO/IEC MPEG-4 AVC), including its Scalable Video Coding (SVC) and Multi-view Video Coding (MVC) extensions. Additionally, a new video coding standard, the High Efficiency Video Coding (HEVC) standard, is currently being developed by the Joint Collaboration Team on Video Coding (JCT-VC) of the ITU-T Video Coding Experts Group (VCEG) and the ISO/IEC Moving Picture Experts Group (MPEG). As of 20 July 2012, the latest Working Draft (WD) of HEVC is available from http://phenix.int-evry.fr/jct/doc_end_user/documents/10_Stockholm/wg11/JCTVC-J1003-v8.zip (and referred to as HEVC WD8 hereinafter).

However, the techniques of this disclosure are not limited to any particular coding standard. For illustrative purposes only, the techniques are described according to the HEVC standard.

Although not shown in FIG1 , in some aspects, video encoder 20 and video decoder 30 may each be integrated with an audio encoder and decoder and may include appropriate MUX-DEMUX units or other hardware and software to handle the encoding of both audio and video in a common data stream or in separate data streams. If applicable, the MUX-DEMUX units may conform to the ITU H.223 multiplexer protocol, or other protocols such as the User Datagram Protocol (UDP).

Video encoder 20 and video decoder 30 may each be implemented as any of a variety of suitable encoder circuits, such as one or more processors (including microprocessors), digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), discrete logic, software, hardware, firmware, or any combination thereof. When the techniques are implemented partially in software, a device may store instructions for the software in a suitable, non-transitory computer-readable medium and execute the instructions in hardware using one or more processors to perform the techniques of this disclosure.

Each of video encoder 20 and video decoder 30 may be included in one or more encoders or decoders, and either of video encoder 20 and video decoder 30 may be integrated as part of a combined encoder/decoder (CODEC) in a respective device. In some examples, video encoder 20 and video decoder 30 may be collectively referred to as video coders that code information (e.g., pictures and syntax elements). When the video coder corresponds to video encoder 20, coding of information may refer to encoding. When the video coder corresponds to video decoder 30, coding of information may refer to decoding.

Furthermore, the techniques described in this disclosure may refer to video encoder 20 signaling information. When video encoder 20 signals information, the techniques of this disclosure generally refer to any means by which video encoder 20 provides the information. For example, when video encoder 20 signals syntax elements to video decoder 30, this may mean that video encoder 20 transmits the syntax elements to video decoder 30 via output interface 24 and communication channel 16, or that video encoder 20 stores the syntax elements on storage medium 17 and/or file server 19 via output interface 24 for eventual reception by video decoder 30. In this manner, signaling from video encoder 20 to video decoder 30 should not be interpreted as requiring transmission from video encoder 20 for immediate reception by video decoder 30, although such a scenario may be possible. Rather, signaling from video encoder 20 to video decoder 30 should be interpreted as any technique by which video encoder 20 provides information for eventual reception by video decoder 30, either directly or via intermediate storage (e.g., in storage medium 17 and/or file server 19).

Video encoder 20 and video decoder 30 may be configured to implement the example techniques described in this disclosure for deriving reference picture sets. For example, video decoder 30 may invoke the process for deriving a reference picture set once per picture. Video decoder 30 may invoke the process for deriving a reference picture set after decoding a slice header but before decoding any coding units and before the decoding process for reference picture list construction for the slice.

As described above, a reference picture set is an absolute description of reference pictures that are used in the decoding process of the current picture and future coded pictures in decoding order until the next instantaneous decoded refresh (IDR) picture or broken link access (BLA) picture. In the examples described in this disclosure, video encoder 20 may explicitly signal a value from which video decoder 30 may determine identifiers for reference pictures belonging to the reference picture set. Reference picture set signaling is explicit in the sense that all reference pictures included in the reference picture set are explicitly listed (except for specific pictures, such as IDR pictures), no reference picture set syntax elements are included in the slice header, and the reference picture set is derived to be empty.

There may be various ways by which video encoder 20 may signal syntax elements in the coded bitstream that video decoder 30 may utilize to derive a reference picture set. For example, video encoder 20 may signal the syntax elements in a picture parameter set (PPS), a sequence parameter set (SPS), a picture header (if present), a slice header, or any combination thereof. For purposes of illustration only, video encoder 20 may signal the syntax elements using an SPS, a PPS, and a slice header, as described in greater detail.

To derive a reference picture set, video decoder 30 may perform a decoding process to determine identifiers for pictures belonging to the reference picture set. Video decoder 30 may then construct multiple reference picture subsets, each of which identifies zero or more of the reference pictures belonging to the reference picture set. Video decoder 30 may derive the reference picture set from the constructed reference picture subsets. For example, video decoder 30 may list the multiple reference picture subsets in a particular order to derive the reference picture set.

There may be various ways by which video decoder 30 can determine identifiers for pictures belonging to a reference picture set. Generally speaking, video encoder 20 may signal a value from which video decoder 30 can determine identifiers for pictures, including pictures belonging to a reference picture set. The identifier for a picture may be a PicOrderCnt (i.e., a picture order number (POC) value). As described above, a POC value may indicate the display or output order of a picture, with pictures with smaller POC values being displayed earlier than pictures with larger POC values. The POC value of a given picture may be relative to a previous instantaneous decode refresh (IDR) picture. For example, the PicOrderCnt (i.e., POC value) of an IDR picture may be 0, a POC value of 1 may be subsequent to the IDR picture in display or output order, a POC value of 2 may be subsequent to the picture with a POC value of 1 in display or output order, and so on.

According to the techniques described in this disclosure, when the current picture is not an IDR picture, the following may be applied to derive the POC value of the current picture. The following is intended to aid understanding and should not be considered limiting.

For example, consider a list variable listD that contains as elements PicOrderCnt values (POC values) associated with a list of pictures that contains all of the following: (1) the first picture in the list is the previous IDR picture in decoding order, and (2) all other pictures that follow the first picture in the list in decoding order and precede the current picture in decoding order or are the current picture. In this example, the current picture is included in listD before the derivation process for the reference picture set is called. Furthermore, consider a list variable listO that contains the elements of listD sorted in ascending order of POC values. In this example, listO may not contain a POC value that is equal to the POC value of another picture.

In some examples, POC values may be limited to a range of -2 ^pocLen-1 to 2 ^{pocLen - 1} -1, inclusive. In this example, pocLen may be equal to long_term_ref_pic_id_len_delta+long_term_ref_pic_id_delta_len_minus4+4. long_term_ref_pic_id_len_delta and long_term_ref_pic_id_delta_len_minus4 may be syntax elements received by video decoder 30 in the coded bitstream as part of picture parameter set syntax, as described in more detail below. As another example, POC values may be limited to a range of ^{-2 31} to 2 ^{31 -1} , inclusive.

As one example, video decoder 30 may receive a pic_order_cnt_lsb syntax element in a coded bitstream (i.e., a bitstream signaled by video encoder 20). The pic_order_cnt_lsb syntax element may specify the picture sequence number modulo MaxPicOrderCntLsb of the coded picture. The length of the pic_order_cnt_lsb syntax element may be log2_max_pic_order_cnt_lsb_minus4+4 bits. The value of pic_order_cnt_lsb may be in the range of 0 to MaxPicOrderCntLsb-1, inclusive. Video decoder 30 may receive the pic_order_cnt_lsb syntax element in the slice header syntax of the current picture to be decoded.

Video decoder 30 may also receive a log2_max_pic_order_cnt_lsb_minus4 syntax element in the coded bitstream signaled by video encoder 20. Video decoder 30 may receive the log2_max_pic_order_cnt_lsb_minus4 syntax element in a sequence parameter set. The value of log2_max_pic_order_cnt_lsb_minus4 may be in the range of 0 to 12, inclusive. The log2_max_pic_order_cnt_lsb_minus4 syntax element may specify the value of the variable MaxPicOrderCntLsb, which video decoder 30 uses in the decoding process for determining the POC value. For example:

MaxPicOrderCntLsb=2 ^{(log2_max_pic_order_cnt_lsb_minus4+4)} .

Based on these received syntax elements, video decoder 30 may determine a POC value for the current picture as follows. For example, video decoder 30 may determine PicOrderCntMsb for the current picture. The POC value for the current picture may be the determined PicOrderCntMsb for the current picture plus the received pic_order_cnt_lsb for the current picture.

In the following, the function PicOrderCnt(picX) is equal to the POC value for picture X. The function DiffPicOrderCnt(picA, picB) is equal to PicOrderCnt(picA) minus PicOrderCnt(picB). In some examples, the coded bitstream may not include data that generates values for DiffPicOrderCnt(picA, picB) that exceed the range of ^{-2 15} to 2 ^{15 -1} (inclusive) for use in the decoding process. Furthermore, assume that X is the current picture and Y and Z are two other pictures in the same sequence, where when DiffPicOrderCnt(X, Y) and DiffPicOrderCnt(X, Z) are both positive or DiffPicOrderCnt(X, Y) and DiffPicOrderCnt(X, Z) are both negative, Y and Z are considered to be in the same output order direction from X. Also, in some examples, video encoder 20 may assign a PicOrderCnt that is proportional to the sampling time of the corresponding picture relative to the sampling time of the previous IDR picture.

As part of the process of determining the POC value for the current picture, video decoder 30 may determine the variables prevPicOrderCntMsb and prevPicOrderCntLsb. For example, if the current picture is an IDR picture, video decoder 30 may set prevPicOrderCntMsb to 0 and prevPicOrderCntLsb to 0. Otherwise (i.e., if the current picture is not an IDR picture), video decoder 30 may set prevPicOrderCntMsb to be equal to the PicOrderCntMsb of the previous reference picture in decoding order that has a temporal_id less than or equal to the temporal_id of the current picture, and set prevPicOrderCntLsb to be equal to the value of the PicOrderCntLsb of the previous reference picture in decoding order that has a temporal_id less than or equal to the temporal_id of the current picture.

With these variable values and the values of syntax elements (e.g., the values of prevPicOrderCntMsb, prevPicOrderCntLsb, pic_order_cnt_lsb, and MaxPicOrderCntLsb), video decoder 30 may determine the value of PicOrderCntMsb based on the steps set forth in the following pseudo-code. It should be understood that video decoder 30 may implement the steps set forth in the following pseudo-code to determine the PicOrderCntMsb for each current picture used to derive the POC value of the current picture.

if((pic_order_cnt_lsb<prevPicOrderCntLsb)&&((prevPicOrderCntLsb-pic_order_cnt_lsb)>=(MaxPicOrderCntLsb/2)))

PicOrderCntMsb=prevPicOrderCntMsb+MaxPicOrderCntLsb

else if((pic_order_cnt_lsb>prevPicOrderCntLsb)&&((pic_order_cnt_lsb-prevPicOrderCntLsb)>(MaxPicOrderCntLsb/2)))

PicOrderCntMsb=prevPicOrderCntMsb-MaxPicOrderCntLsb

else

PicOrderCntMsb=prevPicOrderCntMsb

After determining PicOrderCntMsb for the current picture, video decoder 30 may determine a POC value for the current picture based on PicOrderCntMsb for the current picture and pic_order_cnt_lsb for the current picture. Video decoder 30 may determine the POC value for the current picture as follows:

PicOrderCnt=PicOrderCntMsb+pic_order_cnt_lsb.

After decoding a picture, video decoder 30 may store the PicOrderCntMsb value, pic_order_cnt_lsb value, and POC value for the picture (including each of the reference pictures belonging to a reference picture set) in a decoded picture buffer (DPB) of video decoder 30. In this manner, each picture in the DPB is associated with a POC value, a PicOrderCntMsb value, and a pic_order_cnt_lsb value.

A method for determining POC values for reference pictures included in the reference picture set of the current picture is described in more detail below. Based on the determined POC values, video decoder 30 can implement a derivation process for the reference picture set. However, before describing the manner in which video decoder 30 implements the derivation process for the reference picture set, a table of syntax elements that video decoder 30 can receive in a coded bitstream signaled by video encoder 20 is provided below. For example, video encoder 20 can signal the syntax elements in the following table in the coded bitstream received by video decoder 30. Some of these syntax elements have been described above. Based on these syntax elements, video decoder 30 can determine POC values for reference pictures included in the reference picture set and further derive the reference picture set.

For example, in the techniques described in this disclosure, the following syntax structures are modified relative to previous video coding standards: sequence parameter set (SPS) raw byte sequence payload (RBSP) syntax seq_paramater_set_rbsq(), picture parameter set (PPS) RBSP syntax pic_parameter_set_rbsp(), slice header syntax slice_header(), and reference picture list modification syntax ref_pic_list_modification(). Reference picture list modification is described in more detail following the description of deriving reference picture sets and initializing one or more reference picture lists.

Furthermore, in accordance with the techniques described in this disclosure, the following syntax structures are added to the coded bitstream: a short-term reference picture set syntax, short_term_ref_pic_set(), and a long-term reference picture set syntax, long_term_ref_pic_set(). Video decoder 30 may utilize the short-term reference picture set syntax and the long-term reference picture set syntax for the purpose of constructing reference picture subsets from which it derives a reference picture set.

For example, in order for video decoder 30 to determine a POC value for a reference picture belonging to a reference picture set, video encoder 20 may signal, in a picture parameter set, reference picture identification information that video decoder 30 uses to determine the POC value, and may reference an index to the list in a slice header. However, this is only one example of how video encoder 20 may signal such reference picture identification information.

In one alternative example, video encoder 20 may signal reference picture information in a sequence parameter set and may reference an index to a list in a slice header, which may reduce signaling overhead. In another alternative example, the video coder may signal reference picture information in a new type of parameter set, such as a reference picture set parameter set (RPSPS), and may reference both the RPSPS ID and an index to a list of reference picture identification information in a slice header. This may reduce signaling overhead and not increase the number of picture parameter sets or sequence parameter sets. In other examples, video encoder 20 may utilize any combination of these example techniques to signal reference picture identification information.

Table 1. Sequence parameter set RBSP syntax

pic_width_in_luma_samples may specify the width of each decoded picture in luma samples. The value of ^{pic_width_in_luma_samples} may be in the range of 0 to 2 ¹⁶ −1, inclusive.

pic_height_in_luma_samples may specify the height of each decoded picture in luma samples. The value of ^{pic_height_in_luma_samples} may be in the range of 0 to 2 ¹⁶ −1, inclusive.

As indicated in Table 1, video decoder 30 may receive the log2_max_pic_order_cnt_lsb_minus4 syntax element in a sequence parameter set (SPS). As described above, the value of log2_max_pic_order_cnt_lsb_minus4 may specify the value of the variable MaxPicOrderCntLsb that video decoder 30 uses in the decoding process for determining the POC value, where MaxPicOrderCntLsb=2 ^{(log2_max_pic_order_cnt_lsb_minus4+4)} .

Table 2. Picture parameter set RBSP syntax

num_short_term_ref_pic_sets_pps specifies the number of short_term_ref_pic_set() syntax structures contained in the picture parameter set. The value of num_short_term_ref_pic_sets_pps shall be in the range of 0 to 32, inclusive.

long_term_ref_pics_present_flag equal to 0 specifies that no long-term reference picture is used for inter prediction of any coded picture with reference to the picture parameter set, and the syntax elements long_term_ref_pic_id_delta_len_minus4, long_term_ref_pic_id_len_delta, and num_long_term_ref_pics_pps are not present. long_term_ref_pics_present_flag equal to 1 specifies that a long-term reference picture may be used for inter prediction of one or more coded pictures with reference to the picture parameter set, and the syntax elements long_term_ref_pic_id_delta_len_minus4, long_term_ref_pic_id_len_delta, and num_long_term_ref_pics_pps are present.

long_term_ref_pic_id_delta_len_minus4+4 specifies the length in bits of the long_term_ref_pic_id_delta_add_foll[i] syntax element. The value of long_term_ref_pic_id_delta_len_minus4 shall be in the range of 0 to 12, inclusive.

long_term_ref_pic_id_len_delta + long_term_ref_pic_id_delta_len_minus4+4 specifies the length in bits of the long_term_ref_pic_id_pps[i] syntax element. The value of long_term_ref_pic_id_len_delta can be in the range of 0 to 28 - long_term_ref_pic_id_delta_len_minus4, inclusive. The value of long_term_ref_pic_id_len_delta + long_term_ref_pic_id_delta_len_minus4+4 can be the same in all picture parameter sets that reference a particular sequence parameter set.

num_long_term_ref_pics_pps specifies the number of identified long-term reference pictures included in the picture parameter set. The value of num_long_term_ref_pics_pps may be in the range of 0 to 32, inclusive.

long_term_ref_pic_id_pps[i] specifies the i-th long-term reference picture identification information included in the picture parameter set. The number of bits used to represent long_term_ref_pic_id_pps[i] may be equal to long_term_ref_pic_id_len_delta+long_term_pic_id_len_minus4+4.

Table 3. Short-term reference picture set syntax

Short-term reference picture set syntax may be used for short-term pictures. A short-term picture may be defined as a reference picture whose identification information is included in the short_term_ref_pic_set() syntax structure for the coded picture, which is included in the slice header(s) or in the referenced picture parameter set and referenced by the short_term_ref_pic_set_idx syntax element in the slice header(s). The slice header syntax elements are provided in Table 4 below.

When the short_term_ref_pic_set() syntax structure is used to derive a reference picture set for a coded picture, num_short_term_curr0 specifies the number of short-term reference pictures in RefPicSetStCurr0, as described below. The value of num_short_term_curr0 may be in the range of 0 to max_num_ref_frames, inclusive.

When the short_term_ref_pic_set() syntax structure is used to derive a reference picture set for a coded picture, num_short_term_curr1 specifies the number of short-term reference pictures in RefPicSetStCurr1, as described below. The value of num_short_term_curr1 may be in the range of 0 to max_num_ref_frames-num_short_term_curr0, inclusive.

When the short_term_ref_pic_set() syntax structure is used to derive a reference picture set for a coded picture, num_short_term_foll0 specifies the number of short-term reference pictures in RefPicSetStFoll0, as described below. The value of num_short_term_foll0 may be in the range of 0 to max_num_ref_frames-num_short_term_curr0-num_short_term_curr1, inclusive.

When the short_term_ref_pic_set() syntax structure is used to derive a reference picture set for a coded picture, num_short_term_foll1 specifies the number of short-term reference pictures in RefPicSetStFoll1, as described below. The value of num_short_term_foll1 shall be in the range of 0 to max_num_ref_frames - num_short_term_curr0 - num_short_term_curr1 - num_short_term_foll0, inclusive.

short_term_ref_pic_id_delta_minus1[i] specifies the identification information of the i-th short-term reference picture contained in the short_term_ref_pic_set() syntax structure.

Table 4. Slice header syntax

The no_output_of_prior_pics_flag specifies how previously decoded pictures in the decoded picture buffer are handled after an IDR picture is decoded. When the IDR picture is the first IDR picture in the bitstream, the value of no_output_of_prior_pics_flag may have no effect on the decoding process. When the IDR picture is not the first IDR picture in the bitstream and the values of pic_width_in_luma_samples or pic_height_in_luma_samples or max_dec_frame_buffering derived from the valid sequence parameter set may be different from the values of pic_width_in_luma_samples or pic_height_in_luma_samples or max_dec_frame_buffering derived from the valid sequence parameter set for the previous picture, a no_output_of_prior_pics_flag equal to 1 may (but is not necessarily) be inferred by the decoder, regardless of the actual value of no_output_of_prior_pics_flag.

short_term_ref_pic_set_pps_flag equal to 1 specifies that identification information of the short-term reference picture set contained in the reference picture set for the current picture is present in the referenced picture parameter set. short_term_ref_pic_set_pps_flag equal to 0 specifies that identification information of the short-term reference picture set contained in the reference picture set for the current picture is not present in the referenced picture parameter set.

short_term_ref_pic_set_idx specifies the index of the short_term_ref_pic_set() syntax structure contained in the referenced picture parameter set, which contains identification information of the short-term reference picture set in the reference picture set for the current picture.

Specify the variables NumShortTermCurr0 and NumShortTermCurr1 as:

NumShortTermCurr0=num_short_term_curr0

NumShortTermCurr1=num_short_term_curr1

In the case where num_short_term_curr0 and num_short_term_curr1 are respectively syntax elements of the same name in the short_term_ref_pic_set() syntax structure, they exist in the referenced picture parameter set and are referenced by short_term_ref_pic_set_idx, or they exist directly in the slice header.

num_ref_idx_l0_active_minus1 specifies the maximum reference index for reference picture list 0 to be used to decode the slice.

When the current slice is a P or B slice and num_ref_idx_l0_active_minus1 does not exist, num_ref_idx_l0_active_minus1 may be inferred to be equal to num_ref_idx_l0_default_active_minus1.

The value of num_ref_idx_l0_active_minus1 may be in the range of 0 to 15, inclusive.

num_ref_idx_l1_active_minus1 specifies the maximum reference index for reference picture list 1 to be used to decode the slice.

When the current slice is a P or B slice and num_ref_idx_l0_active_minus1 does not exist, num_ref_idx_l1_active_minus1 may be inferred to be equal to num_ref_idx_l1_default_active_minus1.

The value of num_ref_idx_l1_active_minus1 may be in the range of 0 to 15, inclusive.

Table 5. Long-term reference picture set syntax

Long-term reference picture set syntax may be used for long-term pictures. A long-term picture may be defined as a reference picture whose identification information is contained in the long_term_ref_pic_set() syntax structure for the coded picture.

num_long_term_pps_curr specifies the number of all long-term reference pictures whose identification information is contained in the referenced picture parameter set and which can be used for inter prediction of the current picture. If num_long_term_pps_curr is not present, then the value may be derived to be equal to 0. The value of num_long_term_pps_curr may be in the range of 0 to max_num_ref_frames (inclusive).

num_long_term_add_curr specifies the number of all long-term reference pictures whose identification information is not included in the referenced picture parameter set and that can be used for inter prediction of the current picture. If num_long_term_add_curr is not present, then the value may be derived to be equal to 0. The value of num_long_term_add_curr may be in the range of 0 to max_num_ref_frames - num_long_term_pps_curr (inclusive).

Assign the variable NumLongTermCurr as:

NumLongTermCurr=num_long_term_pps_curr+num_long_term_add_curr

num_long_term_pps_foll specifies the number of all long-term reference pictures whose identification information is contained in the referenced picture parameter set, which are not used for inter prediction of the current picture, and which can be used for inter prediction of any picture following the current picture in decoding order. If num_long_term_pps_foll is not present, then the value may be derived to be equal to 0. The value of num_long_term_pps_foll may be in the range of 0 to max_num_ref_frames (inclusive).

num_long_term_add_foll specifies the number of all long-term reference pictures whose identification information is not included in the referenced picture parameter set, which are not used for inter prediction of the current picture, and which can be used for inter prediction of any subsequent pictures in decoding order. If num_long_term_add_foll is not present, the value may be derived to be equal to 0. The value of num_long_term_add_foll may be in the range of 0 to max_num_ref_frames - num_long_term_pps_foll (inclusive).

long_term_ref_pic_set_idx_pps[i] specifies the index of the list of long-term reference picture identification information contained in the referenced picture parameter set for the i-th long-term reference picture inherited from the referenced picture parameter set to the reference picture set of the current picture. The value of long_term_ref_pic_set_idx_pps[i] can be in the range of 0 to 31 (inclusive).

long_term_ref_pic_id_delta_add[i] specifies the long-term reference picture identification information of the i-th long-term reference picture that is not inherited from the referenced picture parameter set but is included in the reference picture set of the current picture. The number of bits used to represent long_term_ref_pic_id_add_curr[i] may be equal to long_term_pic_id_len_minus4+4.

Using the values signaled or derived above (i.e., the values in Tables 1-5), video decoder 30 may derive a reference picture set. As described above, the derived reference picture set may identify reference pictures that can potentially be used to code/predict the current picture (i.e., the picture currently being decoded) and pictures that follow the current picture in decoding order. According to the techniques described in this disclosure, all reference pictures in the derived reference picture set have a decoding order that is earlier than the decoding order of the current picture.

The derivation process may include constructing a reference picture set from multiple reference picture subsets. This process may be invoked once per picture after decoding the slice header but before decoding any coding units and before the decoding process for constructing the reference picture list for the slice. For example, based on the derived values and signaled syntax elements, video decoder 30 may determine POC values for reference pictures belonging to the reference picture set. Based on the determined POC values, video decoder 30 may construct a subset of reference pictures that together form a reference picture set. In this manner, by constructing the reference picture subsets, video decoder 30 may construct a reference picture set. For example, video decoder 30 may order the reference picture subsets in a particular manner to derive the reference picture set. Ordering may refer to the manner in which video decoder 30 lists the reference picture subsets to derive the reference picture set.

As described above, to derive a reference picture set, video decoder 30 may construct multiple reference picture subsets. In some examples, video decoder 30 may construct six reference picture subsets. The six reference picture subsets may be named: RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, RefPicSetStFoll1, RefPicSetLtCurr, and RefPicSetLtFoll. RefPicSetStCurr0 may be referred to as RefPicSetStCurrBefore, and RefPicSetStCurr1 may be referred to as RefPicSetStCurrAfter.

It should be understood that six reference picture subsets are described for illustrative purposes and should not be construed as limiting. As an example, video decoder 30 may construct fewer than six reference picture subsets, for example, by combining some of the subsets. Some of these examples in which video decoder 30 constructs fewer than six reference picture subsets are described below. However, for illustrative purposes, the techniques are described using an example in which video decoder 30 constructs six reference picture subsets.

The RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, and RefPicSetStFoll1 reference picture subsets may identify short-term reference pictures. In some examples, these reference picture subsets may identify short-term reference pictures based on whether the short-term reference picture is earlier in display order than the current picture being coded or later in display order than the current picture being coded, and whether the short-term reference picture can potentially be used for inter-prediction of the current picture and pictures that follow the current picture in decoding order, or can potentially be used for inter-prediction of only pictures that follow the current picture in decoding order.

For example, the RefPicSetStCurr0 reference picture subset may include (and may only include) identification information (e.g., POC values) of all short-term reference pictures that have an output or display order earlier than that of the current picture and that are potentially usable for reference in inter-frame prediction of the current picture and that are potentially usable for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order. The RefPicSetStCurr1 reference picture subset may include (and may only include) identification information of all short-term reference pictures that have an output or display order later than that of the current picture and that are potentially usable for reference in inter-frame prediction of the current picture and that are potentially usable for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order.

The RefPicSetStFoll0 reference picture subset may include (and may only include) identification information of all short-term reference pictures that satisfy the following conditions: have an output or display order earlier than that of the current picture, may be potentially used for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order, and cannot be used for reference in inter-frame prediction of the current picture. The RefPicSetStFoll1 reference picture subset may include (and may only include) identification information of all short-term reference pictures that satisfy the following conditions: have an output or display order later than that of the current picture, may be potentially used for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order, and cannot be used for reference in inter-frame prediction of the current picture.

The RefPicSetLtCurr and RefPicSetLtFoll reference picture subsets may identify long-term reference pictures. In some examples, these reference picture subsets may identify long-term reference pictures based on whether the long-term reference picture is earlier in display order than the current picture being coded or later in display order than the current picture being coded.

For example, the RefPicSetLtCurr reference picture subset may include (and may only include) identification information of all long-term reference pictures that can potentially be used for reference in inter-frame prediction of the current picture and can potentially be used for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order. The RefPicSetLtFoll reference picture subset may include (and may only include) identification information of all long-term reference pictures that can potentially be used for reference in inter-frame prediction of one or more pictures that follow the current picture in decoding order and cannot be used for reference in inter-frame prediction of the current picture.

If the current picture to be decoded is an IDR picture, video decoder 30 may set the RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, RefPicSetStFoll1, RefPicSetLtCurr, and RefPicSetLtFoll reference picture subsets to empty. This may be because IDR pictures cannot be inter-predicted and pictures following the IDR picture in decoding order cannot use any pictures preceding the IDR picture for reference during decoding. Otherwise (e.g., when the current picture is a non-IDR picture), video decoder 30 may construct the short-term reference picture subset and the long-term reference picture subset by implementing the following pseudo code.

For example, when video decoder 30 decodes an instance of the short_term_ref_pic_set() syntax structure in a slice header or referenced picture parameter set, video decoder 30 may implement the following pseudocode to construct the RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, and RefPicSetStFoll1 reference picture subsets.

cIdx=0

for(i＝0,pocPred＝PicOrderCnt(CurrPic); i<num_short_term_curr0; pocPred＝RefPicSetStCurr0[i],i++,cIdx++)

RefPicSetStCurr0[i]=pocPred-short_term_ref_pic_id_delta_minus1[cIdx]-1

for(i＝0,pocPred＝PicOrderCnt(CurrPic); i<num_short_term_curr1; pocPred＝RefPicSetStCurr1[i],i++,cIdx++)

RefPicSetStCurr1[i]=short_term_ref_pic_id_delta_minus1[cIdx]+1-pocPred

for(i＝0,pocPred＝PicOrderCnt(CurrPic); i<num_short_term_foll0; pocPred＝RefPicSetStFoll[i],i++,cIdx++)

RefPicSetStFoll0[i]=pocPred-short_term_ref_pic_id_delta_minus1[cIdx]-1

for(i＝0,pocPred＝PicOrderCnt(CurrPic); i<num_short_term_foll1; pocPred＝RefPicSetStFoll[i],i++,cIdx++)

RefPicSetStFoll1[i]=short_term_ref_pic_id_delta_minus1[cIdx]+1-pocPred

If video decoder 30 determines that long_term_ref_pics_present_flag is equal to 0, then video decoder 30 may set RefPicSetLtCurr and RefPicSetLtFoll to null because no long-term reference pictures exist for this case. Otherwise, if video decoder 30 decodes an instance of the long_term_ref_pic_set() syntax structure in a slice header, then video decoder 30 may implement the following pseudo-code to construct the RefPicSetLtCurr and RefPicSetLtFoll reference picture subsets.

According to the techniques described in this disclosure, a reference picture with a particular value of PicOrderCnt (POC value) is said to be included in a reference picture set for a coded picture if the reference picture is included in any of the six subsets of the reference picture set for the coded picture. A reference picture with a particular value of PicOrderCnt is said to be included in a particular subset of the reference picture set if the particular value of PicOrderCnt (POC value) is equal to one of the PicOrderCnt values included in the subset.

After constructing the reference picture subsets, video decoder 30 may derive the reference picture sets. For example, video decoder 30 may order the reference picture subsets to derive the reference picture sets. As an example, video decoder 30 may list the RefPicSetStCurr0 reference picture subset, followed by the RefPicSetStCurr1 reference picture subset, followed by the RefPicSetStFoll0 reference picture subset, followed by the RefPicSetStFoll1 reference picture subset, followed by the RefPicSetLtCurr reference picture subset, and then the RefPicSetLtFoll reference picture subset. As another example, video decoder 30 may list the RefPicSetStCurr0 reference picture subset, followed by the RefPicSetStCurr1 reference picture subset, followed by the RefPicSetLtCurr reference picture subset, followed by the RefPicSetStFoll0 reference picture subset, followed by RefPicSetStFoll1, and then the RefPicSetLtFoll reference picture subset.

Other permutations of the manner in which video decoder 30 orders the reference picture subsets may be possible for deriving the reference picture set. In some examples, the construction of the reference picture subsets may be combined with the derivation of the reference picture set. For example, the construction of the reference picture subsets may automatically cause video decoder 30 to derive the reference picture set. In other words, video decoder 30 may not need to perform different steps for constructing the reference picture subsets and deriving the reference picture set, but it may be possible for video decoder 30 to first construct the reference picture subsets and then derive the reference picture set.

Furthermore, constructing the reference picture set in the manner described above, according to the techniques described in this disclosure, may result in video decoder 30 satisfying the following constraints. For example, a particular reference picture having a particular value of PicOrderCnt may not be included in more than one of the reference picture subsets in the reference picture set of the current picture. In other words, a reference picture identified in one of the reference picture subsets may not be identified in any of the other reference picture subsets. As another example, in the derived reference picture set, there may not be a reference picture having a temporal_id greater than the temporal_id of the current picture included in any of the reference picture subsets forming the reference picture set.

As described above, a temporal identification value (temporal_id) may be a hierarchical value that indicates which pictures can be used for coding/predicting the current picture. Generally speaking, a picture with a particular temporal_id value may be a reference picture for pictures with equal or greater temporal_id values, but the reverse is not true. For example, a picture with a temporal_id value of 1 may be a reference picture for pictures with temporal_id values of 1, 2, 3, ..., but not for a picture with a temporal_id value of 0.

The lowest temporal_id value may also indicate the lowest display rate. For example, if video decoder 30 only decodes pictures with a temporal_id value of 0, the display rate may be 7.5 pictures/second. If video decoder 30 only decodes pictures with temporal_id values 0 and 1, the display rate may be 15 pictures/second, and so on.

In some examples, only pictures with temporal_id values less than or equal to the temporal_id of the current picture can be included in the reference picture set for the current picture. As described above, only pictures with temporal_id values less than or equal to the temporal_id of the current picture can be used as reference pictures. Therefore, all reference pictures with lower or equal temporal_id values can be used by the current picture for inter-frame prediction and can be included in the reference picture set. Furthermore, some reference pictures are excluded from the reference picture set if they have temporal_id values greater than the temporal_id value of the current picture, or if they are to be used by pictures that follow the current picture in decoding order and have temporal_id values greater than the temporal_id value of the current picture.

Through these techniques, signaling of temporal_id and picture identification for deriving a reference picture set is not required; therefore, reference picture set signaling becomes more efficient. For example, video encoder 20 may not signal temporal_id values for reference pictures belonging to a reference picture set, and video decoder 30 may not need to receive temporal_id values for reference pictures belonging to a reference picture set for the purpose of deriving a reference picture set.

Furthermore, in this manner, the constructed reference picture subsets may not identify reference pictures having temporal_id values greater than the temporal_id value of the current picture. For example, video decoder 30 may be able to construct the reference picture subsets and ensure that no reference picture identified in any of the reference picture subsets has a temporal_id value greater than the temporal_id value of the current picture, because bitstream consistency may require that the temporal_id value not be included in the bitstream signaled by video encoder 20 and received by video decoder 30. In this manner, video decoder 30 may derive a reference picture set without receiving temporal identification values for reference pictures belonging to the reference picture set.

In the above example, video decoder 30 may construct six reference picture subsets, four of which are for short-term reference pictures (i.e., RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, and RefPicSetStFoll1) and two of which are for long-term reference pictures (i.e., RefPicSetLtCurr and RefPicSetLtFoll). However, aspects of this disclosure should not be so limited. In other examples, two or more of these reference picture subsets may be combined into one reference picture subset, resulting in video decoder 30 constructing fewer reference picture subsets. Some non-limiting examples of how video decoder 30 may construct fewer reference picture subsets are described below. There may be other ways in which video decoder 30 may construct fewer reference picture subsets.

For example, in some instances, there may not be a separation between the subset for the current picture and the subset for subsequent pictures in decoding order. Thus, there may be two subsets for short-term reference pictures (referred to as RefPicSetSt0 and RefPicSetSt1), and only one subset for long-term reference pictures (referred to as RefPicSetLt). In this example, the RefPicSetSt0 reference picture subset may be the concatenation of RefPicSetStCurr0 and RefPicSetStFoll0, with RefPicSetStCurr0 at the beginning of the concatenation. In this example, the RefPicSetSt1 reference picture subset may be the concatenation of RefPicSetStCurr1 and RefPicSetStFoll1, with RefPicSetStCurr1 at the beginning of the concatenation. The RefPicSetLt reference picture subset may be the concatenation of RefPicSetLtCurr and RefPicSetLtFoll, with RefPicSetLtCurr at the beginning of the concatenation.

As another example, there may not be a separation of subsets having an output order earlier or later than the current picture. This scenario may only apply to short-term reference pictures. Thus, there may be two subsets for short-term reference pictures, referred to as RefPicSetStCurr and RefPicSetStFoll. The RefPicSetStCurr reference picture subset may be the concatenation of RefPicSetStCurr0 and RefPicSetStCurr1, with RefPicSetStCurr0 at the beginning of the concatenation. The RefPicSetStFoll reference picture subset may be the concatenation of RefPicSetStFoll0 and RefPicSetStFoll1, with RefPicSetStFoll0 at the beginning of the concatenation.

As another example, the two types of separation described above may not apply. Thus, there may be only one subset for short-term reference pictures (called RefPicSetSt) and only one subset for long-term reference pictures (called RefPicSetLt). The RefPicSetSt reference picture subset may be the concatenation of RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, and RefPicSetStFoll1 in the order listed (or any other order), and RefPicSetLt may be the same as above.

The above techniques describe example ways in which video decoder 30 may derive a reference picture set. During the encoding process, video encoder 20 may also need to decode encoded pictures for the purpose of encoding subsequent pictures, a process known as a reconstruction process. Therefore, in some examples, video encoder 20 may also be configured to derive a reference picture set. In some examples, video encoder 20 may implement the same techniques as video decoder 30 to derive a reference picture set. However, it may not be necessary for video encoder 20 to derive a reference picture set in every example, and video decoder 30 may be the only coder deriving a reference picture set.

Thus, in some examples, a video coder (e.g., video encoder 20 or video decoder 30) may code (e.g., encode or decode, respectively) information indicating reference images that belong to a reference image set. For example, video encoder 20 may signal an encoded bitstream that includes values used to determine which reference images belong to the reference image set. Similarly, video decoder 30 may decode the bitstream to determine which reference images belong to the reference image set.

The video coder may construct multiple reference picture subsets, each reference picture subset identifying zero or more of the reference pictures. For example, the video coder may construct six reference picture subsets (i.e., RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, RefPicSetStFoll1, RefPicSetLtCurr, and RefPicSetLtFoll reference picture subsets), each of which identifies zero or more reference pictures. In some examples, the video coder may code a current picture based on the multiple reference picture subsets.

For example, the video coder may derive a reference picture set from the constructed multiple reference picture subsets. For example, the video coder may sort the reference picture subsets in any order to derive the reference picture set, or may derive the reference picture set as part of constructing the reference picture subsets. In some examples, the video coder may code the current picture based on the derived reference picture set. Because the reference picture set is derived from the multiple reference picture subsets, the video coder may be considered to be coding the current picture based on the multiple reference picture subsets.

In some examples, to order the reference picture subsets, the video coder may construct the reference picture subsets in the order in which they are to be listed in the reference picture set. For example, the video coder may first construct the RefPicSetLtCurr reference picture subset, then construct the RefPicSetLtFoll reference picture subset, then construct the RefPicSetStCurr0 reference picture subset, then construct the RefPicSetStCurr1 reference picture subset, then construct the RefPicSetStFoll0 reference picture subset, and then construct the RefPicSetStFoll1 reference picture subset. In this illustrative example, the order of the reference picture subsets in the reference picture set may be RefPicSetLtCurr, RefPicSetLtFoll, RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, and RefPicSetStFoll1 (in that order), although other orders are possible.

According to the example techniques described in this disclosure, after deriving a reference picture set, video decoder 30 can decode a slice within the current picture. Part of the decoding process involves constructing one or two reference picture lists. A reference picture list is a list of reference pictures used for prediction of a P or B slice. For the decoding process of a P slice, there is one reference picture list (List 0). For the decoding process of a B slice, there are two reference picture lists (List 0 and List 1). List 0 (sometimes referred to as Reference Picture List 0 or RefPicList0) is the reference picture list used for inter-frame prediction of a P or B slice. All inter-frame predictions for P slices use List 0. Reference Picture List 0 is one of two reference picture lists used for bidirectional prediction of B slices, with the other reference picture list being Reference Picture List 1. List 1 (sometimes referred to as Reference Picture List 1 or RefPicList1) is the reference picture list used for prediction of a B slice. Reference Picture List 1 is one of two reference picture lists used for prediction of a B slice, with the other reference picture list being Reference Picture List 0. Some blocks in a B slice may be bi-directionally predicted using both list 0 and list 1, and some blocks in a B slice may be uni-directionally predicted using either list 0 or list 1.

To construct the reference picture lists, video decoder 30 may implement a default construction technique to construct Initial List 0 and (for B slices) Initial List 1. The construction of Initial List 0 and Initial List 1 may be referred to as an initialization process. In some examples, the coded bitstream may indicate that video decoder 30 should modify Initial List 0 and/or Initial List 1 to produce Final List 0 and Final List 1. The modification of Initial List 0 and/or Initial List 1 may be referred to as a modification process. The modification process may not be required in every example, and the manner in which video decoder 30 may implement the modification process is described in more detail below. According to the techniques described in this disclosure, when modification of Initial List 0 or Initial List 1 is not required, Final List 0 or Final List 1 (i.e., the reference picture list 0 or 1 used to decode the slice of the current picture) may be equal to Initial List 0 or Initial List 1. In this way, reference picture list reordering may not be required.

In the techniques described in this disclosure, video decoder 30 may construct initial List 0 or initial List 1 in a manner that, regardless of whether a modification process is required, video decoder 30 may not need to perform reordering of reference pictures to be included in initial List 0 or initial List 1 because the reference pictures in each of the reference picture subsets are already in the proper order. For example, in some other techniques, regardless of whether a modification process is required, when adding or listing a reference picture to initial List 0 or initial List 1, the reference pictures to be included in initial List 0 or initial List 1 need to be reordered according to their POC values.

During initialization, video decoder 30 may implement a default construction technique to construct initial list 0 and initial list 1. The default construction technique may mean that video decoder 30 constructs the initial reference picture lists without receiving syntax elements from video encoder 20 regarding the manner in which video decoder 30 should construct the initial reference picture lists or which reference pictures should be identified in the initial reference picture lists.

When decoding a P or B slice header, video decoder 30 may invoke a reference picture list construction process. For example, when decoding a P slice, video decoder 30 may invoke a process for constructing Initial List 0, but may not invoke a process for constructing Initial List 1, because blocks in a P slice are only unidirectionally predicted with respect to the reference pictures identified in List 0. When decoding a B slice, video decoder 30 may invoke a process for constructing Initial List 0 and a process for constructing Initial List 1, because blocks in a B slice may be bidirectionally predicted with respect to the reference pictures identified in each of List 0 and List 1.

According to example techniques described in this disclosure, video decoder 30 may construct Initial List 0 and Initial List 1 using subsets of reference pictures. For example, Initial List 0 and Initial List 1 may list zero or more reference pictures identified in RefPicSetStCurr0, RefPicSetStCurr1, or RefPicSetLtCurr. In this example, at least one reference picture may be present in RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr when the reference picture list construction process is called. Although Initial List 0 and Initial List 1 may identify one or more reference pictures from the same subset of reference pictures, the order in which video decoder 30 adds the reference pictures in Initial List 0 may be different from the order in which video decoder 30 adds the reference pictures in Initial List 1.

In this disclosure, when video decoder 30 adds (e.g., lists) one or more reference pictures from a reference picture subset in initial List 0 or initial List 1, this disclosure refers to video decoder 30 identifying the reference pictures in initial List 0 or initial List 1. For example, the multiple reference picture subsets may each identify zero or more reference pictures. To construct initial List 0 and initial List 1, video decoder 30 may identify one or more of the reference pictures identified in the reference picture subset to be included in initial List 0 or initial List 1.

To avoid confusion and to aid clarity, this disclosure may refer to video decoder 30 listing or adding zero or more of the reference pictures identified in the reference picture subset to Initial List 0 and Initial List 1 to construct Initial List 0 and Initial List 1. In this manner, when video decoder 30 adds or lists reference pictures, it means that video decoder 30 adds or lists the identifiers of the reference pictures identified in the reference picture subset. Thus, the resulting Initial List 0 and Initial List 1 contain multiple identifiers for reference pictures that can potentially be used to decode a block or slice of the current picture. These reference pictures are stored in the respective decoded picture buffers of video decoder 30 and video encoder 20.

For example, to construct initial list 0, video decoder 30 may first list (e.g., add) the reference pictures identified in RefPicSetStCurr0 in initial list 0, followed by listing (e.g., adding) the reference pictures identified in RefPicSetStCurr1 in initial list 0, and then list (e.g., add) the reference pictures identified in RefPicSetLtCurr in initial list 0. To construct initial list 1, video decoder 30 may first list (e.g., add) the reference pictures identified in RefPicSetStCurr1 in initial list 1, followed by listing (e.g., adding) the reference pictures identified in RefPicSetStCurr0 in initial list 1, and then list (e.g., add) the reference pictures identified in RefPicSetLtCurr in initial list 1.

Furthermore, in addition to adding the reference pictures in the reference picture subsets in a different order, video decoder 30 may also utilize a different number of reference pictures from each of the reference picture subsets when constructing List 0 and List 1. For example, List 0 and List 1 need not include all reference pictures from RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr. Instead, the number of reference pictures listed from these example reference picture subsets to construct Initial List 0 and Initial List 1 may be based on a syntax element indicating the maximum number of reference pictures within each of Initial List 0 and Initial List 1.

For example, for initial list 0, video encoder 20 may signal in the slice header a num_ref_idx_l0_active_minus1 syntax element for P and B slices, and a num_ref_idx_l1_active_minus1 syntax element for bi-predicted B slices. As described above, num_ref_idx_l0_active_minus1 may define the maximum number of reference pictures that may be in initial list 0, and num_ref_idx_l1_active_minus1 may define the maximum number of reference pictures that may be in initial list 1. In some examples, it may be possible that the value of num_ref_idx_l0_active_minus1 is different from the value of num_ref_idx_l1_active_minus1, but this may not be necessary in every example. In some examples, the value of num_ref_idx_l0_active_minus1 may be the same as the value of num_ref_idx_l1_active_minus1.

As described above, video decoder 30 may receive values for num_short_term_curr0 and num_short_term_curr1 in the coded bitstream. Video decoder 30 may define the variable NumShortTermCurr0 equal to num_short_term_curr0 and the variable NumShortTermCurr1 equal to num_short_term_curr1. NumShortTermCurr0 may indicate the number of reference pictures in the RefPicSetStCurr0 reference picture subset, and NumShortTermCurr1 may indicate the number of reference pictures in the RefPicSetStCurr1 reference picture subset.

Video decoder 30 may also receive values for num_long_term_pps_curr and num_long_term_add_curr in the coded bitstream. Video decoder 30 may define the variable NumLongTermCurr equal to num_long_term_pps_curr plus num_long_term_add_curr. NumLongTermCurr may indicate the number of reference pictures in RefPicSetLtCurr.

To construct initial List 0, video decoder 30 may first add reference pictures in RefPicSetStCurr0 to initial List 0, until video decoder 30 has added all reference pictures in RefPicSetStCurr0 to initial List 0, and as long as the number of entries in initial List 0 (e.g., the number of reference pictures identified in List 0) is less than or equal to num_ref_idx_10_active_minus1. For example, NumShortTermCurr0 may indicate the number of reference pictures in the reference picture subset of RefPicSetStCurr0. In this example, video decoder 30 may list (e.g., add) reference pictures from the reference picture subset of RefPicSetStCurr0 until the number of reference pictures listed from RefPicSetStCurr0 is equal to NumShortTermCurr0. However, when listing the reference pictures of RefPicSetStCurr0 into initial List 0, if the total number of entries in initial List 0 is equal to num_ref_idx_10_active_minus1, video decoder 30 may stop adding reference pictures in the RefPicSetStCurr0 reference picture subset, even if there are additional pictures in RefPicSetStCurr0 that are not listed in initial List 0. In this case, video decoder 30 may have completed construction of initial List 0.

After video decoder 30 lists all reference pictures in the RefPicSetStCurr0 reference picture subset and the total number of entries in initial List 0 is less than num_ref_idx_l0_active_minus1, video decoder 30 may then add reference pictures in RefPicSetStCurr1 until video decoder 30 identifies all reference pictures in RefPicSetStCurr1, as long as the number of entries in initial List 0 (e.g., the number of reference pictures identified in List 0) is less than or equal to num_ref_idx_l0_active_minus1. For example, similar to the above scenario, NumShortTermCurr1 may indicate the number of reference pictures in the RefPicSetStCurr1 reference picture subset. In this example, video decoder 30 may list reference pictures from the RefPicSetStCurr1 reference picture subset until the number of reference pictures listed from RefPicSetStCurr1 is equal to NumShortTermCurr1. However, when listing reference pictures from RefPicSetStCurr1, if the total number of entries in initial list 0 is equal to num_ref_idx_10_active_minus1, video decoder 30 may stop adding reference pictures from the RefPicSetStCurr1 reference picture subset, even if there are additional pictures in RefPicSetStCurr1 that are not listed in initial list 0. In this case, video decoder 30 may have completed construction of initial list 0.

After video decoder 30 lists all reference pictures in the RefPicSetStCurr1 reference picture subset and the total number of entries in initial list 0 is less than num_ref_idx_l0_active_minus1, video decoder 30 may then list reference pictures in RefPicSetLtCurr until video decoder 30 lists all reference pictures in RefPicSetLtCurr, as long as the number of entries in initial list 0 (e.g., the number of reference pictures identified in list 0) is less than or equal to num_ref_idx_l0_active_minus1. For example, similar to the above scenario, NumLongTermCurr may indicate the number of reference pictures in the RefPicSetLtCurr reference picture subset. In this example, video decoder 30 may list reference pictures from the RefPicSetLtCurr reference picture subset until the number of reference pictures listed from RefPicSetLtCurr is equal to NumLongTermCurr. However, when listing reference pictures from RefPicSetLtCurr into initial List 0, if the total number of entries in initial List 0 is equal to num_ref_idx_10_active_minus1, video decoder 30 may stop adding reference pictures in the reference picture subset of RefPicSetLtCurr, even if there are additional pictures in RefPicSetLtCurr that are not listed in initial List 0. In this case, video decoder 30 may have completed construction of initial List 0.

The following pseudo-code illustrates how video decoder 30 may construct initial list 0.

In the above pseudo-code, RefPicList0 may be the initial list 0. In instances where modification of list 0 is not required, the final list 0 may be equal to the initial list 0. Therefore, in instances where modification of list 0 is not required, in the above pseudo-code, RefPicList0 may be the final list 0.

Video decoder 30 may similarly construct initial list 1. However, to construct initial list 1, video decoder 30 may first add reference pictures from the RefPicSetStCurr1 reference picture subset to initial list 1, followed by adding reference pictures from the RefPicSetStCurr0 reference picture subset to initial list 1, and then followed by adding reference pictures from the RefPicSetLtCurr reference picture subset to initial list 1. Also, similar to the above scenario, when listing reference pictures from any of the RefPicSetStCurr1, RefPicSetStCurr0, and RefPicSetLtCurr reference picture subsets, if the total number of entries in initial list 1 is equal to num_ref_idx_l1_active_minus1, video decoder 30 may stop adding reference pictures, even if additional reference pictures exist in these reference picture subsets.

For example, to construct Initial List 1, video decoder 30 may first list reference pictures from RefPicSetStCurr1 until video decoder 30 identifies all reference pictures in RefPicSetStCurr1, and as long as the number of entries in Initial List 1 (e.g., the number of reference pictures identified in List 1) is less than or equal to num_ref_idx_l1_active_minus1. For example, the value of NumShortTermCurr1 may indicate when video decoder 30 has completed listing all reference pictures in the reference picture subset of RefPicSetStCurr1. However, while listing reference pictures from RefPicSetStCurr1, if the total number of entries in Initial List 1 is equal to num_ref_idx_l1_active_minus1, video decoder 30 may stop adding reference pictures in the reference picture subset of RefPicSetStCurr1, even if there are additional pictures in RefPicSetStCurr1 that are not listed in Initial List 1. In this case, video decoder 30 may have completed construction of Initial List 1.

After video decoder 30 lists all reference pictures in the RefPicSetStCurr1 reference picture subset and the total number of entries in initial list 1 is less than num_ref_idx_l1_active_minus1, video decoder 30 may then list reference pictures from RefPicSetStCurr0 until video decoder 30 lists all reference pictures in RefPicSetStCurr0, and as long as the number of entries in initial list 1 (e.g., the number of reference pictures identified in list 1) is less than or equal to num_ref_idx_l1_active_minus1. For example, similar to the scenario described above, the value of NumShortTermCurr0 may indicate when video decoder 30 has finished listing all reference pictures in the RefPicSetStCurr0 reference picture subset. However, when listing reference pictures from RefPicSetStCurr0 into initial List 1, if the total number of entries in initial List 1 is equal to num_ref_idx_l1_active_minus1, video decoder 30 may stop adding reference pictures in the RefPicSetStCurr0 reference picture subset, even if there are additional pictures in RefPicSetStCurr0 that are not listed in initial List 1. In this case, video decoder 30 may have completed construction of initial List 1.

After video decoder 30 lists all reference pictures in the RefPicSetStCurr0 reference picture subset and the total number of entries in initial list 1 is less than num_ref_idx_l1_active_minus1, video decoder 30 may then list reference pictures in RefPicSetLtCurr until video decoder 30 lists all reference pictures in RefPicSetLtCurr, and as long as the number of entries in initial list 1 (e.g., the number of reference pictures identified in list 1) is less than or equal to num_ref_idx_l1_active_minus1. For example, similar to the scenario described above, the value of NumLongTermCurr may indicate when video decoder 30 has finished listing all reference pictures in the RefPicSetLtCurr reference picture subset. However, when listing reference pictures from RefPicSetLtCurr, if the total number of entries in initial List 1 is equal to num_ref_idx_l1_active_minus1, video decoder 30 may stop adding reference pictures in the RefPicSetLtCurr reference picture subset, even if there are additional pictures in RefPicSetLtCurr that are not listed in initial List 1. In this case, video decoder 30 may have completed construction of initial List 1.

The following pseudo-code illustrates how video decoder 30 may construct initial list 1.

In the above pseudo-code, RefPicList1 may be the initial list 1. In instances where modification of list 1 is not required, final list 1 may be equal to initial list 1. Therefore, in instances where modification of list 1 is not required, in the above pseudo-code, RefPicList1 may be the final list 1.

The foregoing example approach is one example approach by which video decoder 30 may construct Final List 0 and Final List 1 when reference picture list modification is not required. In other examples, video decoder 30 may add reference picture subsets in an order different from the order described above. In still other examples, video decoder 30 may add reference picture subsets different from the reference picture subsets described above.

Although the foregoing examples describe techniques for reference picture list construction performed by video decoder 30, aspects of this disclosure are not so limited, and video encoder 20 may implement similar techniques for constructing reference picture lists. However, video encoder 20 may not necessarily construct the reference picture lists in the same manner as video decoder 30 does.

Thus, a video coder (e.g., video encoder 20 or video decoder 30) may be configured to code (e.g., encode or decode) information indicating reference pictures that belong to a reference picture set. As described above, a reference picture set identifies reference pictures that may potentially be used for inter-predicting a current picture and that may potentially be used for inter-predicting one or more pictures that follow the current picture in decoding order.

The video coder may also be configured to construct multiple reference picture subsets, each reference picture subset identifying zero or more reference pictures. For example, the video coder may construct at least the RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets. The video coder may construct additional reference picture subsets, such as the reference picture subsets described above.

As long as the number of initial reference picture list entries is not greater than the maximum number of allowable reference picture list entries, the video coder may then add the following reference pictures to the initial reference picture list: a reference picture from the first reference picture subset, followed by a reference picture from the second reference picture subset, followed by a reference picture from the third reference picture subset. For example, as long as the number of entries in initial List 0 is not greater than num_ref_idx_l0_active_minus1, the video coder may include the following reference pictures in initial List 0: a reference picture from the RefPicSetStCurr0 reference picture subset, followed by a reference picture from the RefPicSetStCurr1 reference picture subset, followed by a reference picture from the RefPicSetLtCurr subset. Furthermore, the value of num_ref_idx_l0_active_minus1 may indicate the maximum number of allowable reference picture list entries for List 0.

In some examples, the video coder may add reference pictures from the first reference picture subset to the initial reference picture list until all reference pictures from the first reference picture subset are listed in the initial reference picture list or the number of initial reference picture list entries equals the maximum number of allowable reference picture list entries. When the number of initial reference picture list entries is less than the maximum number of allowable reference picture list entries, and after adding the reference pictures from the first reference picture subset, the video coder may add reference pictures from the second reference picture subset to the initial reference picture list until all reference pictures from the second reference picture subset are listed in the initial reference picture list or the number of initial reference picture list entries equals the maximum number of allowable reference picture list entries. When the number of initial reference picture list entries is less than the maximum number of allowable reference picture list entries, and after adding the reference pictures from the second reference picture subset, the video coder may add reference pictures from the third reference picture subset to the initial reference picture list until all reference pictures from the third reference picture subset are listed in the initial reference picture list or the number of initial reference picture list entries equals the maximum number of allowable reference picture list entries.

The video coder may similarly construct Initial List 1. For example, as long as the number of initial reference picture list entries in Initial List 1 is not greater than num_ref_idx_l1_active_minus1, the video coder may add the following reference pictures to Initial List 1: a reference picture from the second reference picture subset, followed by a reference picture from the first reference picture subset, and followed by a reference picture from the third reference picture subset. The num_ref_idx_l1_active_minus1 syntax element may define the maximum number of entries allowed in List 1.

In some examples, e.g., when no modification is required, Initial List 0 and Initial List 1 may be equal to Final List 0 and Final List 1. In other words, when no modification is required, the video coder may construct Final List 0 and Final List 1 without modifying Initial List 0 and Initial List 1. In these cases, after constructing Initial List 0 and Initial List 1, the video coder may not need to perform additional steps to construct Final List 0 and Final List 1 (i.e., reference picture lists used by the video coder to code blocks of the current picture).

As indicated in the pseudocode above, video decoder 30 may construct Initial List 0 when cIdx is less than or equal to num_ref_idx_l0_active_minus1, and may construct Initial List 1 when cIdx is less than or equal to num_ref_idx_l1_active_minus1. This may result in video decoder 30 constructing Initial List 0 and Initial List 1 without any incomplete entries in the reference picture lists. For example, in some other video coding technologies, the video decoder for these other video technologies will construct Initial List 0 and List 1 using techniques different from those described in this disclosure. For these other video coding technologies, if the number of entries in Initial List 0 and Initial List 1 is less than the maximum allowable number of entries, the video decoder for these other video coding technologies will fill the remaining entries in List 0 and List 1 with "no reference picture" entries that are incomplete. Incomplete entries refer to entries in List 0 and List 1 after the last entry that identifies a reference picture and up to the last possible entry.

As an illustrative example to aid understanding, a video decoder for these other video coding techniques may construct List 0 with five entries, where the maximum number of allowable entries is ten. In this example, the video decoder for these other video coding techniques fills the sixth through tenth entries with "no reference picture." In this example, the incomplete entries would be the sixth entry (e.g., the entry after the last entry identifying a reference picture) through the tenth entry (e.g., the last possible entry as defined by the maximum number of allowable entries).

According to the techniques of this disclosure, video decoder 30 may construct Initial List 0 and Initial List 1 so that there are no incomplete entries. Furthermore, in instances where reference picture list modification is not required, Final List 0 and Final List 1 may be equal to Initial List 0 and Initial List 1. Thus, in instances where reference picture list modification is not required, video decoder 30 may construct Final List 0 and Final List 1 so that there are no incomplete entries. Even in instances where reference picture list modification is required, the modification may not result in any incomplete entries. Thus, even in instances where reference picture list modification is required, video decoder 30 may construct Final List 0 and Final List 1 so that there are no incomplete entries.

For example, List 0 and List 1 may be considered as lists with entries, and each entry may identify a reference picture (e.g., by the POC value of the reference picture). In other words, video decoder 30 may identify the reference picture by the POC value of the reference picture in each entry of List 0 and List 1. The number of entries in List 0 and List 1 may be defined by the num_ref_idx_l0_active_minus1 and num_ref_idx_l1_active_minus1 syntax elements, respectively.

To ensure that there are no incomplete entries, video decoder 30 may repeatedly list (e.g., add or identify) reference pictures from the reference picture subset in initial List 0 and initial List 1 until video decoder 30 determines which reference picture should be identified in each possible entry of initial List 0 and initial List 1. For example, as described above, for constructing initial List 0, after adding reference pictures from the RefPicSetStCurr0 and RefPicSetStCurr1 reference picture subsets in initial List 0, video decoder 30 may add reference pictures from the RefPicSetLtCurr reference picture subset in initial List 0.

In some examples, the following scenario may be possible: after video decoder 30 adds reference pictures from the RefPicSetLtCurr reference picture subset in initial List 0, the total number of entries in initial List 0 is less than the maximum number of allowable entries in List 0. For example, in the pseudo code for constructing initial List 0, cIdx may indicate the number of entries in List 0. In some examples, after video decoder 30 identifies reference pictures in RefPicSetLtCurr in initial List 0, the value of cIdx may be less than num_ref_idx_l0_active_minus1, where num_ref_idx_l0_active_minus1 specifies the maximum number of allowable entries in List 0.

According to the techniques described in this disclosure, after listing reference pictures from three of the multiple reference picture subsets, if the number of entries in initial List 0 is less than the maximum number of allowable entries, video decoder 30 may repeatedly add reference pictures from the three reference picture subsets until all entries in List 0 are full. For example, after video decoder 30 adds reference pictures from the RefPicSetLtCurr reference picture subset and the number of entries in initial List 0 is less than the maximum number of allowable entries, video decoder 30 may then re-list (e.g., re-add or re-identify) reference pictures from the RefPicSetStCurr0 reference picture subset.

In aspects described in this disclosure, when video decoder 30 lists reference pictures from RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets, video decoder 30 may be considered to be adding the reference pictures from these multiple reference picture subsets to a first set of entries in a reference picture list (e.g., List 0). For example, the first set of entries may be entries in the reference picture list in which video decoder 30 identifies reference pictures from the RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets. Then, if the number of entries in the reference picture list is less than the maximum allowable number of entries, video decoder 30 may re-list (e.g., re-add or re-identify) one or more reference pictures from at least one of the RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets to entries in the reference picture list that follow the first set of entries. Entries subsequent to the first set of entries may be entries following the first set of entries in which video decoder 30 adds already listed reference pictures from one or more of the RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets, as described below.

When re-adding reference pictures from the RefPicSetStCurr0 reference picture subset, if the total number of entries in initial list 0 is equal to num_ref_idx_l0_active_minus1, video decoder 30 may stop re-adding reference pictures in initial list 0. In this case, video decoder 30 may have completed construction of initial list 0 and there may be no incomplete entries. Otherwise, video decoder 30 may re-add reference pictures from the RefPicSetStCurr0 reference picture subset until all reference pictures from the RefPicSetStCurr0 reference picture subset are re-listed.

If, after re-adding all reference pictures in the RefPicSetStCurr0 reference picture subset, the number of entries in initial List0 is less than num_ref_idx_l0_active_minus1, video decoder 30 may then re-add reference pictures from the RefPicSetStCurr1 reference picture subset in a manner similar to the manner by which video decoder 30 re-lists reference pictures from the RefPicSetStCurr0 reference picture subset. If, after re-adding all reference pictures in the RefPicSetStCurr1 reference picture subset, the number of entries in initial List0 is less than num_ref_idx_l0_active_minus1, video decoder 30 may then re-add reference pictures from the RefPicSetLtCurr reference picture subset in a manner similar to the manner by which video decoder 30 re-lists reference pictures from the RefPicSetStCurr0 and RefPicSetStCurr1 reference picture subsets. Video decoder 30 repeatedly adds reference pictures from the reference picture subset until the number of entries in initial List 0 is equal to the maximum number of allowable entries for List 0 (ie, equal to num_ref_idx_l0_active_minus1).

[0066] For example, assume there is one reference picture in RefPicSetStCurr0, one reference picture in RefPicSetCurr1, and one reference picture in RefPicSetLtCurr. Also, assume num_ref_idx_l0_active_minus1 is equal to five. In this example, video decoder 30 may identify a reference picture in RefPicSetStCurr0 in two entries in initial list 0. For example, video decoder 30 may identify a reference picture in RefPicSetStCurr0 in the first entry of initial list 0 and re-identify the reference picture in RefPicSetStCurr0 in the fourth entry of initial list 0. In this example, the index values for the reference picture in RefPicSetStCurr0 may be index[0] for the first entry in initial list 0 and index[3] for the fourth entry in initial list 0. Thus, in some examples, one reference picture from one of the reference picture subsets may be listed (e.g., identified) more than once in the initial reference picture list.

Video decoder 30 may similarly construct initial list 1 such that there are no non-complete entries in initial list 1. For example, video decoder 30 may repeatedly add reference pictures from RefPicSetStCurr1, RefPicSetStCurr0, and RefPicSetLtCurr reference picture subsets (in that order) until the number of entries in initial list 1 equals the maximum number of allowable entries in list 1 (i.e., equal to num_ref_idx_l1_active_minus1).

In this manner, because the "for" loop is nested within the "while" loop, in the above pseudocode for constructing Initial List 0 and Initial List 1, video decoder 30 may construct Initial List 0 and Initial List 1 such that there are no non-complete entries in Initial List 0 and Initial List 1 (i.e., there are no non-complete entries after the initialization process). In some examples, each of the entries in Initial List 0 and Initial List 1 may identify a reference picture from one of the reference picture subsets. In some examples, it may be possible that one or more reference pictures from one of the reference picture subsets are identified more than once in the final reference picture list, but at different entries with different index values.

Although the foregoing examples describe techniques for reference picture list construction without non-completion entries as performed by video decoder 30, aspects of this disclosure are not so limited, and video encoder 20 may implement similar techniques for constructing a reference picture list without non-completion entries. However, video encoder 20 may not necessarily construct the reference picture list in the same manner as video decoder 30 constructs the reference picture list.

Thus, a video coder (e.g., video encoder 20 or video decoder 30) may be configured to code (e.g., encode or decode) information indicating reference pictures that belong to a reference picture set. As described above, a reference picture set identifies reference pictures that may potentially be used for inter-predicting a current picture and that may potentially be used for inter-predicting one or more pictures that follow the current picture in decoding order.

A video coder may construct multiple reference picture subsets, each of which identifies zero or more of the reference pictures (e.g., RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets). The video coder may list (e.g., identify or add) the reference pictures from the multiple reference picture subsets in a first set of entries in a reference picture list. The video coder may determine whether the number of entries in the reference picture list is equal to a maximum number of allowable entries in the reference picture list.

When the number of entries in the reference picture list is not equal to the maximum number of allowable entries in the reference picture list, the video coder may repeatedly re-add (e.g., re-identify) one or more reference pictures from at least one of the reference picture subsets to entries following the first set of entries in the reference picture list until the number of entries in the reference picture list is equal to the maximum number of allowable entries in the reference picture list. The video coder may then decode the current picture based on the constructed reference picture list.

As described above, in some examples, video encoder 20 may signal syntax elements that direct video decoder 30 to modify one or more initial reference picture lists. For example, video decoder 30 may construct one or more initial reference picture lists in the manner described above. Then, in some examples, video decoder 30 may decode syntax elements from the coded bitstream that direct video decoder 30 to modify the one or more initial reference picture lists to construct one or more final reference picture lists. Generally speaking, in modifying, after initializing the reference picture lists, video decoder 30 may map one or more of the pictures identified in one or more of the plurality of reference picture subsets to an entry in one of the reference picture lists.

For example, after video decoder 30 constructs one or more initial reference picture lists in the manner described above, video decoder 30 may modify at least one of the entries in one of the initial reference picture lists, as directed by the coded bitstream. For example, video encoder 20 may indicate which picture from one of the plurality of reference picture subsets should be identified in an entry of a reference picture list (as part of a modification syntax element), even though the entry of the reference picture list may already have identified a reference picture (as part of an initialization process). In some examples, the reference picture list modification techniques described in this disclosure may allow for modifications to be made in a flexible manner. For example, video decoder 30 may be able to identify a reference picture in one or both of the reference picture lists that is not in the initial reference picture list.

As used in this disclosure, the phrase "modified reference picture list" refers to a reference picture list after the initial reference picture list has been modified. The modified reference picture list may be the final reference picture list. For List 0, the number of entries in the modified reference picture list is num_ref_idx_l0_active_minus1+1, and for List 1, the number of entries in the modified reference picture list is num_ref_idx_l1_active_minus1+1. A reference picture may be present at more than one index (e.g., entry) in the modified reference lists for both List 0 and List 1.

For reference picture list modification, video encoder 20 may signal the syntax elements of Table 6.

Table 6. Reference picture list modification syntax

The syntax elements of modification_of_ref_pic_idc, short_term_ref_pic_set_idx, and long_term_ref_pic_set_idx may specify the change from an initial reference picture list to a reference picture list to be used for decoding a slice.

ref_pic_list_modification_flag_10 equal to 1 may specify that the syntax element modification_of_ref_pic_idc is present to specify reference picture list 0. ref_pic_list_modification_flag_10 equal to 0 specifies that this syntax element is not present.

When ref_pic_list_modification_flag_10 is equal to 1, the number of times modification_of_ref_pic_idc is not equal to 3 following ref_pic_list_modification_flag_10 may not exceed num_ref_idx_10_active_minus1+1.

ref_pic_list_modification_flag_l1 equal to 1 may specify that the syntax modification_of_ref_pic_idc is present for specifying reference picture list 1. ref_pic_list_modification_flag_l1 equal to 0 may specify that this syntax element is not present.

When ref_pic_list_modification_flag_l1 is equal to 1, the number of times modification_of_ref_pic_idc is not equal to 3 following ref_pic_list_modification_flag_l1 may not exceed num_ref_idx_l1_active_minus1+1.

modification_of_ref_pic_idc together with short_term_ref_pic_set_idx or long_term_ref_pic_set_idx may specify which of the reference pictures are remapped. The values of modification_of_ref_pic_idc are specified in Table 7. The value of the first modification_of_ref_pic_idc immediately following ref_pic_list_modification_flag_l0 or ref_pic_list_modification_flag_l1 may not be equal to 3.

ref_pic_set_idx specifies the index into RefPicSetStCurr0, RefPicSetStCurr1, or RefPicSetLtCurr of the reference picture to be moved to the current index in the reference picture list. The value of ref_pic_set_idx may be in the range of 0 to max_num_ref_frames (inclusive).

Table 7. modification_of_ref_pic_idc operation for modification of reference picture list

For reference picture list modification, when ref_pic_list_modification_flag_l0 is equal to 1, video decoder 30 may modify the initial reference picture list 0 (i.e., initial list 0), and when ref_pic_list_modification_flag_l1 is equal to 1, video decoder 30 may modify the initial reference picture list 1 (i.e., initial list 1). To aid in understanding reference picture list modification, assume that the variable refIdxL0 is an index into initial list 0, and the variable refIdxL1 is an index into initial list 1. In other words, refIdxL0 may identify an entry of initial list 0 (i.e., an index into initial list 0 identifies an entry of initial list 0), and refIdxL1 may identify an entry of initial list 1. Initially, the variables refIdxL0 and refIdxL1 may be set equal to 0.

Video decoder 30 may process the syntax elements of modification_of_ref_pic_idc in the order in which they appear in the bitstream. For example, if video encoder 20 signals that initial list 0 requires reference picture list modification, video decoder 30 may process the modification_of_ref_pic_idc syntax element signaled by video encoder 20 for modifying the order of initial list 0. Similarly, if video encoder 20 signals that initial list 1 requires reference picture list modification, video decoder 30 may process the modification_of_ref_pic_idc syntax element signaled by video encoder 20 for modifying the order of initial list 1.

The value of the modification_of_ref_pic_idc syntax element may be 0, 1, 2, or 3, as indicated in Table 7. If the value of the modification_of_ref_pic_idc syntax element is 3, video decoder 30 may stop modification of the initial reference picture list. Otherwise, video decoder 30 may keep modifying the initial reference picture list until the value of the modification_of_ref_pic_idc syntax element is 3. For example, video encoder 20 may signal multiple values for the modification_of_ref_pic_idc syntax element, and video decoder 30 may process each of the values in the order in which they appear in the coded bitstream. When video decoder 30 processes a value of the modification_of_ref_pic_idc syntax element of 3, video decoder 30 may determine that no further modification is required.

A value of the modification_of_ref_pic_idc syntax element slightly different from 3 (i.e., 0, 1, or 2) may indicate from which subset of reference pictures video decoder 30 is to identify a reference picture to be included in (e.g., added to) the current entry of the reference picture list. As described above, the current entry of the reference picture list can be identified by the value of refIdxLX, where LX is List 0 or List 1. For example, if video decoder 30 is modifying initial List 0 and modification_of_ref_pic_idc is 0, then according to Table 7, video decoder 30 may determine which reference picture from RefPicSetStCurr0 is to be identified in the current entry of the reference picture list based on the value of ref_pic_set_idx. If video decoder 30 is modifying initial List 1 and modification_of_ref_pic_idc is 0, then according to Table 7, video decoder 30 may determine which reference picture from RefPicSetStCurr1 is to be identified in the current entry of the reference picture list based on the value of ref_pic_set_idx. For example, the variable curRefPicSet may define which subset of reference pictures video decoder 30 should use to modify initial list 0 or initial list 1.

For example, if modification_of_ref_pic_idc is equal to 0, and video decoder 30 is modifying initial List 0, then curRefPicSet is equal to RefPicSetStCurr0 reference picture subset. If modification_of_ref_pic_idx is equal to 0, and video decoder 30 is modifying initial List 1, then curRefPicSet is equal to RefPicSetStCurr1 reference picture subset.

If video decoder 30 is modifying initial list 0, and modification_of_ref_pic_idc is 1, then according to Table 7, video decoder 30 may determine which reference picture from RefPicSetStCurr1 to identify in the current entry of the reference picture list based on the value of ref_pic_set_idx. If video decoder 30 is modifying initial list 1, and modification_of_ref_pic_idc is 1, then according to Table 7, video decoder 30 may determine which reference picture from RefPicSetStCurr0 to identify in the current entry of the reference picture list based on the value of ref_pic_set_idx.

In this case, if modification_of_ref_pic_idc is equal to 1, and video decoder 30 is modifying Initial List 0, then curRefPicSet is equal to the RefPicSetStCurr1 reference picture subset. If modification_of_ref_pic_idx is equal to 1, and video decoder 30 is modifying Initial List 1, then curRefPicSet is equal to the RefPicSetStCurr0 reference picture subset.

If video decoder 30 is modifying Initial List 0 or Initial List 1, and modification_of_ref_pic_idc is 2, then according to Table 7, video decoder 30 may determine which reference picture from RefPicSetLtCurr to identify in the current entry of the reference picture list based on the value of ref_pic_set_idx. In this example, if modification_of_ref_pic_idc is equal to 2, and video decoder 30 is modifying Initial List 0 or Initial List 1, then curRefPicSet is equal to the RefPicSetLtCurr reference picture subset.

As described above, the ref_pic_set_idx syntax element may indicate an index into one of the multiple reference picture subsets. In other words, the ref_pic_set_idx syntax element may indicate an entry from one of the multiple reference picture subsets to video decoder 30. Video decoder 30 may determine that the reference picture identified in the entry in one of the multiple reference picture subsets is the reference picture to be identified in the current index of initial list 0 or initial list 1.

The variable curRefPicPoc may be equal to PicOrderCnt(curRefPicSet[ref_pic_set_idx]). In this manner, the value of curRefPicPoc may be the POC value of the reference picture identified in the ref_pic_set_idx entry of curRefPicSet. As described above, curRefPicSet may be equal to RefPicSetStCurr0, RefPicSetStCurr1, or RefPicSetLtCurr based on the value of the modification_of_ref_pic_idc syntax element and based on whether video decoder 30 is modifying Initial List 0 or Initial List 1.

Video decoder 30 may implement the following pseudocode for reference picture list modification. For example, in the following pseudocode, video decoder 30 may identify a picture in an entry of an initial reference picture list that has a POC value equal to curRefPicPoc. The variable refIdxLX indicates the index position for the entry in the initial reference picture list. For example, when video decoder 30 is modifying initial list 0, refIdxLX may be refIdxL0, and when video decoder 30 is modifying initial list 1, refIdxLX may be refIdxL1.

After the video decoder 30 identifies a reference picture having a POC value equal to curRefPicPOC in the initial reference picture list (i.e., initial list 0 or initial list 1), the video decoder 30 may shift the positions of the other remaining pictures to later positions in the list. For example, the video decoder 30 may move the reference picture identified in the entry following the current entry in the initial reference picture list to the next entry to construct a modified reference picture list. As an illustrative example, assume that the current entry in the initial reference picture list is the third entry with index [2]. The video decoder 30 may move the reference picture currently identified in the third entry with index [2] to the next entry (e.g., the fourth entry with index [3]). The video decoder 30 may move the reference picture currently identified in the fourth entry with index [3] to the fifth entry with index [4]. In some instances, the video decoder 30 may start with the last entry in the initial reference picture list and move the reference pictures identified in that entry to a temporary new entry. The reference picture identified in the second to last entry is then moved to the last entry, and so on, until video decoder 30 reaches the current entry.

Video decoder 30 may then increment the value of variable refIdxLX. In this pseudo-code, the length of RefPicListX (i.e., RefPicList0 or RefPicList1) is made one element longer in time than the required length of the final reference picture list. For example, as described above, video decoder 30 may start with the last entry in the initial reference picture list, move the last entry to the temporary entry, move the derivative second entry to the last entry, and so on, to modify the initial reference picture list. After executing the pseudo-code, video decoder 30 may retain only the entries in index 0 to num_ref_idx_lX_active_minus1, where for list 0, num_ref_idx_lX_active_minus1 is num_ref_idx_l0_active_minus1, and for list 1, num_ref_idx_lX_active_minus1 is num_ref_idx_l1_active_minus1.

In the above pseudocode, RefPicListX refers to either RefPicList0 (i.e., final List 0) or RefPicList1 (i.e., final List 1) based on whether video decoder 30 is modifying initial List 0 or initial List 1. num_ref_idx_lX_active_minus1 refers to either num_ref_idx_l0_active_minus1 or ref_idx_l1_active_minus1 based on whether video decoder 30 is modifying initial List 0 or initial List 1.

The above techniques describe example ways in which video decoder 30 may modify initial reference picture lists. During the encoding process, video encoder 20 may also need to decode encoded pictures for the purpose of encoding subsequent pictures. Therefore, in some examples, video encoder 20 may also be configured to construct initial reference picture lists and modify the initial reference picture lists in the manner described above. However, video encoder 20 may not need to modify one or more initial reference picture lists in every example. In some examples, video decoder 30 may be the only coder that modifies initial reference picture lists using the techniques described above.

Thus, in some examples, a video coder (e.g., video encoder 20 or video decoder 30) may construct an initial reference picture list (e.g., initial list 0 or initial list 1) utilizing the techniques described above. The video coder may determine whether reference picture list modification is needed based on coded syntax elements in the coded bitstream. When reference picture list modification is needed, the video coder may modify the initial reference picture list.

For example, when reference picture list modification is required, the video coder may identify a reference picture in at least one of the constructed reference picture subsets. The video coder may list (e.g., add) the identified reference picture subset in the current entry of the initial reference picture list to construct a modified reference picture list. The video coder may then code (e.g., encode or decode) the current picture based on the modified reference picture list.

As an example, a video coder may construct RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets. To identify a reference picture in at least one of these reference picture subsets, the video coder may determine an index into at least one of these reference picture subsets. The video coder may then determine the reference picture identified at an entry in at least one of these reference picture subsets based on the determined index.

For example, a video coder may code (e.g., encode or decode) a first syntax element (e.g., a modification_of_ref_pic_idc syntax element) by which the video coder may identify one of the reference picture subsets (e.g., one of the RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets). The video coder may code a second syntax element (e.g., a ref_pic_set_idx syntax element) that indicates an index into the identified reference picture subset (e.g., one of the RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets).

In some examples, the video coder may be further configured to move a reference picture in the initial reference picture list. For example, the video coder may move a reference picture identified in an entry following the current entry in the initial reference picture list to the next entry in the modified reference picture list.

The previous examples describe how video encoder 20 and video decoder 30 can derive reference picture sets, as well as example techniques for constructing reference picture lists when modification is not required and when modification is required. However, the techniques described in this disclosure should not be limited thereto. In some examples, the techniques described in this disclosure may be directed to decoded picture buffer (DPB) management. A DPB may be a buffer that stores decoded pictures.

Each of the video encoder 20 and the video decoder 30 may include a respective DPB. For example, as part of the encoding process, the video encoder 20 may decode a current picture, store the decoded picture in the DPB of the video encoder 20, and use the decoded picture stored in the DPB to inter-predict subsequent pictures. Similarly, as part of the decoding process, the video decoder 30 may decode a current picture and store the decoded picture in the DPB of the video decoder 30. The video decoder 30 may then use the decoded picture to inter-predict subsequent pictures.

In some examples, the DPB for video encoder 20 or video decoder 30 may store decoded pictures for output reordering or output delay. For example, video decoder 30 may determine that decoded pictures should be reordered for output or that output of decoded pictures should be delayed. In these examples, the DPB of video decoder 30 may store decoded pictures for output reordering or output delay.

The DPB management techniques described in this disclosure may relate to how the DPB outputs and removes decoded pictures. The output_flag syntax element may affect the decoded picture output and removal process and may be defined as part of the Network Abstraction Layer (NAL) unit semantics. A NAL unit may be defined as a syntax structure that includes an indication of the type of data to follow and bytes containing the data in the form of a Raw Byte Sequence Payload (RBSP), interspersed with contention prevention bytes as necessary. An RBSP may be a syntax structure that includes an integer number of bytes encapsulated in a NAL unit. An RBSP may be empty or in the form of a data bit string consisting of a syntax element followed by an RBSP stop bit and followed by zero or more subsequent bits equal to zero. Table 8 defines the NAL unit syntax.

Table 8. NAL unit syntax

In Table 8, output_flag can affect the decoded picture output and removal process, as described in more detail below. For any picture, if output_flag is equal to 1, then the picture is intended for output. Otherwise, the picture is never output. In the techniques described in this disclosure, the variable OutputFlag is equal to the output_flag syntax element.

In some examples, any coded slice NAL unit of a coded picture of the current access unit may differ from any coded slice NAL unit of a coded picture of the previous access unit in one or more of the following ways. For example, the pic_parameter_set_id value may be different, the nal_ref_idc value may be different, where one of the nal_ref_idc values is equal to 0. The pic_order_cnt_lsb value may be different. The idrPicFlag value may be different. The idrPicFlag may be equal to 1 (for both cases), and the idr_pic_id value may be different.

In the techniques described in this disclosure, an access unit may be defined as a set of NAL units that are consecutive in decoding order and contain one coded picture. An auxiliary coded picture or other NAL unit may not contain slices of a coded picture in addition to the coded picture. In some examples, decoding of an access unit may produce a decoded picture. A coded picture may be a coded representation of a picture to be used by the decoding process.

As indicated in Table 4, the slice header syntax may include a pic_parameter_set_id syntax element, a pic_order_cnt_lsb syntax element, an IdrPicFlag syntax element, and an idr_pic_id syntax element. As indicated in Table 8, the NAL unit syntax may include a nal_ref_idc syntax element.

For purposes of illustration, DPB management techniques are described from the perspective of a hypothetical reference decoder (HRD). An HRD may be defined as a hypothetical decoder model that specifies constraints on the variability of a conforming NAL unit stream or a conforming byte stream that an encoding process may produce. However, according to the techniques described in this disclosure, video decoder 30 may implement DPB management techniques, and in some examples, video encoder 20 may also implement DPB management techniques.

The HDR model may define a coded picture buffer (CPB), a transient decoding process, and a decoded picture buffer (DPB). The CPB may be similar to the CPB of the HDR model defined in other previous standards (i.e., the CPB may store coded pictures). The techniques described in this disclosure relate to DPB operations that differ from those in other standards. Furthermore, it should be understood that video decoder 30, and possibly video encoder 20, may implement DPB operations, as described below.

In general, the techniques described in this disclosure relate to the output and removal of decoded images from the DPB. In this context, output of a decoded image means outputting the decoded image for display, storage, or other purposes. However, the output decoded image need not necessarily be removed from the DPB. For example, the video decoder 30 may not remove the output decoded image from the DPB because the video decoder 30 may need to use the decoded image as a reference image for inter-frame prediction of subsequent images. In this context, removal of a decoded image means removing the decoded image from the DPB.

For example, video decoder 30 may store decoded pictures in the DPB of video decoder 30 in the order in which they were decoded. However, the decoding order of pictures may not be the same as the output order of pictures. For example, there may be pictures that follow the current picture in decoding order and will be output before the current picture. Therefore, in some examples, video decoder 30 may perform reordering, whereby video decoder 30 reorders the pictures in the DPB that were sorted in decoding order into output order. Video decoder 30 may then output the decoded pictures in the output order of the decoded pictures. Video decoder 30 may also remove pictures from the decoded pictures if they do not need to be output (i.e., the pictures have already been output or the pictures are not intended for output) and do not need to be inter-predicted (i.e., do not need to be used as reference pictures for inter-prediction).

In the techniques described in this disclosure, if a decoded picture has already been output or is not intended for output, and if the decoded picture is not identified in the derived reference picture set (which is equivalent to no longer being needed for inter-prediction reference (i.e., the decoded picture no longer needs to be used as a reference picture for inter-prediction)), then video decoder 30 may remove the decoded picture from the DPB. Furthermore, as described above, the reference picture set may identify reference pictures that can potentially be used for inter-prediction of the current picture and for inter-prediction of one or more pictures that follow the current picture in decoding order. According to the techniques described in this disclosure, if a decoded picture is not identified in the derived reference picture set, then the decoded picture may not be needed as a reference picture for inter-prediction (e.g., decoding) of the current picture and one or more pictures that follow the current picture in decoding order. Therefore, if a decoded picture is not intended for output, it may be removed from the DPB because it may not be necessary to retain the decoded picture in the DPB given that the decoded picture will not be used for inter-prediction.

Furthermore, in the techniques described in this disclosure, video decoder 30 may remove decoded images before decoding a current image. For example, as described above, before decoding a current image, video decoder 30 may derive a reference image set and construct reference image list(s). Because video decoder 30 may derive a reference image set before decoding a current image, video decoder 30 may be configured to determine whether decoded images that are not required for output should be removed before decoding the current image. For example, after deriving the reference image set and before decoding the current image, video decoder 30 may determine whether decoded images to be output or decoded images that are not desired for output are not identified in the reference image set. Then, if the decoded images are not identified in the reference image set, video decoder 30 may remove decoded images that are not required for output (i.e., already output or not desired for output) before decoding the current image.

In some examples, video decoder 30 may remove the decoded picture before decoding the current picture. However, video decoder 30 may remove the decoded picture after constructing the reference picture list(s). For example, video decoder 30 may derive a reference picture set and may construct the reference picture list based on the reference picture set. Then, before decoding the current picture, video decoder 30 may remove the decoded picture. In some examples, video decoder 30 may also output the decoded picture after constructing the reference picture list(s).

This disclosure describes techniques for removing decoded pictures from the DPB from at least two perspectives. In the first perspective, if the picture is intended for output, the video decoder 30 can remove the decoded picture based on the output time. In the second perspective, if the picture is intended for output, the video decoder 30 can remove the decoded picture based on the POC value. In either perspective, when the decoded picture is not in the reference picture set and before decoding the current picture, the video decoder 30 can remove the decoded picture that is not required for output (i.e., has already been output or is not intended for output).

The DPB may include multiple buffers, and each buffer may store a decoded picture to be used as a reference picture or maintained for future output. Initially, the DPB is empty (i.e., the DPB fullness is set to zero). In the described example techniques, removing the decoded picture from the DPB may occur before decoding the current picture, but after video decoder 30 analyzes the slice header of the first slice of the current picture.

In a first view, the following techniques may occur instantaneously at time _tr (n) in the following sequence. In this example, _tr (n) is the CPB removal time (i.e., decoding time) of access unit n containing the current picture. As described in this disclosure, the techniques occurring instantaneously may mean that, in the HDR model, it is assumed that decoding of pictures is instantaneous (i.e., infinitely fast), with a time period for decoding a picture equal to zero.

If the current picture is an IDR picture, and when the IDR picture is not the first IDR picture and the values of pic_width_in_luma_samples or pic_height_in_luma_samples, or max_dec_frame_buffering derived from the valid sequence parameter set are different from the values of pic_width_in_luma_samples or pic_height_in_luma_samples, or max_dec_frame_buffering, respectively, derived from the valid sequence parameter set for the previous picture, video decoder 30 may infer that the no_output_of_prior_pics_flag syntax element is equal to 1, regardless of the actual value of no_output_of_prior_pics_flag. If the current picture is an IDR picture, and when no_output_of_prior_pics_flag is equal to 1 or is inferred to be equal to 1, video decoder 30 may flush all buffers of the DPB without outputting pictures in the DPB, and may set the DPB fullness to 0.

As indicated above in Table 1, the sequence parameter set may include the pic_width_in_luma_samples and pic_height_in_luma_samples syntax elements. The sequence parameter set may also include the max_dec_frame_buffering syntax element. As indicated above in Table 4, the slice header syntax may include the no_output_of_prior_pics_flag syntax element.

When the current picture is not an IDR picture, video decoder 30 may remove all pictures (m) in the DPB for which the following conditions hold. The first condition may be that the picture is not included in the reference picture set of the current picture. The second condition may be that the picture has an OutputFlag equal to 0 or the DPB output time of the picture is less than or equal to the CPB removal time of the current picture. In this example, the CPB removal time is t _r (n), which is the instance when the removal process occurs (e.g., the time before the current picture is decoded). The DPB output time of decoded picture m may be defined by the variable t _o,dpb (m). Thus, a DPB output time that is less than or equal to the CPB removal time may be expressed as t _o,dpb (m) < t _r (n). The derivation of the DPB output time (t _o,dpb ) is defined in more detail below.

In this manner, based on the output time of a decoded picture and when the decoded picture is not identified in the reference picture set, video decoder 30 may remove a decoded picture from the DPB before decoding the picture. When video decoder 30 removes a decoded picture from the DPB, video decoder 30 may decrement the DPB fullness by one.

The following describes how video decoder 30 may determine when to output a decoded picture (e.g., the DPB output time of the decoded picture), and also describes when video decoder 30 may store a decoded picture in the DPB. As described above, the DPB output time of a picture may be a factor in determining whether to remove the picture from the DPB.

When video decoder 30 decodes a picture, video decoder 30 stores the picture in the DPB and increments the DPB fullness by 1. When a picture has OutputFlag equal to 1, video decoder 30 may derive the DPB output time for the picture based on the following equation.

t _o,dpb (n)＝t _r (n)+t _c *dpb_output_delay(n)

In the equation, dpb_output_delay(n) may be specified in a picture timing SEI message associated with an access unit including a picture. SEI messages may be defined in some standards, such as the H.264/AVC standard.

The t _o,dpb (n) value may define when to output a picture. For example, if OutputFlag is equal to 1 and t _o,dpb (n) is equal to t _r (n), video decoder 30 may output the picture. Otherwise, if OutputFlag is equal to 0, video decoder 30 may not output the picture. In instances where OutputFlag is equal to 1 and t _o,dpb (n) is greater than t _r (n), video decoder 30 may output the picture at a later time, e.g., at time t _o,dpb (n).

In some examples, when video decoder 30 outputs a picture, video decoder 30 may clip the picture. For example, video decoder 30 may utilize a clipping rectangle specified in a valid sequence parameter set for the picture. Techniques for clipping pictures are generally well established and described in standards such as the H.264/AVC standard.

In some examples, video decoder 30 may determine the difference between the DPB output time of a picture and the DPB output time of pictures that follow the picture in output order. For example, when picture (n) is a picture output by video decoder 30 and is not the last picture in the output bitstream, video decoder 30 may determine the value of Δt _o,dpb (n) to be defined as:

Δt _o,dpb (n)＝t _o,dpb (n _n )-t _o,dpb (n)

In the above equation, n _n indicates a picture that follows picture (n) in output order and has OutputFlag equal to 1. Also, in the above equation, Δt _o,dpb (n) represents the DPB output time difference between a picture and the subsequent picture in output order.

In the second approach to removing decoded pictures, HDR can implement the techniques instantaneously when removing an access unit from the CPB. Furthermore, video decoder 30 can implement the removal of decoded pictures from the DPB, and video decoder 30 may not necessarily include a CPB. Generally speaking, in this disclosure, the removal of decoded pictures is performed by video decoder 30 and may also be performed by video encoder 20. In these examples, video decoder 30 and video encoder 20 may not require a CPB. Instead, the CPB is described as part of the HDR model for illustrative purposes only.

As described above, in the second aspect for removing decoded pictures, video decoder 30 may remove the picture from the DPB before decoding the current picture but after analyzing the slice header of the first slice of the current picture. Furthermore, similar to the first aspect for removing decoded pictures, in the second aspect, video decoder 30 may perform functions similar to those described above with respect to the first aspect when the current picture is an IDR picture.

Otherwise, if the current picture is not an IDR picture, the video decoder 30 may flush the buffer of the DPB without outputting it, the buffer storing pictures marked as "not required for output" and storing pictures not included in the reference picture set of the current picture. The video decoder 30 may also decrement the DPB fullness by the number of buffers flushed by the video decoder 30. When there are no empty buffers (i.e., the DPB fullness is equal to the DPB size), the video decoder 30 may implement the "bumping" process described below. In some examples, when there are no empty buffers, the video decoder 30 may repeatedly implement the bumping process until there is an empty buffer in which the video decoder 30 can store the current decoded picture.

When the current picture is an IDR picture with no_output_of_prior_pics_flag not equal to 1 and not inferred to be equal to 1, video decoder 30 may perform the following operations. Video decoder 30 may flush the buffers of the DPB without outputting the pictures that are marked as "not required for output" and are not included in the reference picture set of the current picture. Video decoder 30 may flush all non-empty buffers in the DPB by repeatedly calling the "bump" process and may set the DPB fullness to 0.

In other words, when the current picture is an IDR picture, video decoder 30 may implement techniques to flush all buffers in the DPB. When the current picture is not an IDR picture, video decoder 30 may implement techniques to remove decoded pictures to free buffers for storing the current decoded picture.

For example, after video decoder 30 decodes the current picture, video decoder 30 may store the current picture in the DPB and increment the DPB fullness by one. In some examples, if the OutputFlag of the current picture is equal to 1, video decoder 30 may mark the current picture as "needed to be output." Otherwise, if the OutputFlag of the current picture is equal to 0, video decoder 30 may mark the current picture as "not needed to be output."

As described above, in some examples, video decoder 30 may implement a bump process. Generally, the bump process involves outputting decoded pictures. For example, when the current picture is an IDR picture and no_output_of_prior_pics_flag is not equal to 1 and has not been inferred to be equal to 1, video decoder 30 may implement a bump process. Video decoder 30 may also implement a bump process if there is no empty buffer in the DPB (i.e., the DPB fullness is equal to the size of the DPB) and an empty buffer is needed for storing decoded (non-IDR) pictures.

In general, the video decoder 30 may implement the following steps to implement the collision process. The video decoder 30 may first determine the picture to be output. For example, the video decoder 30 may select the picture with the smaller PicOrderCnt (POC) value among all the pictures marked as "need to be output" in the DPB. The video decoder 30 may clip the selected picture using the clipping rectangle specified in the valid sequence parameter set for the picture. The video decoder 30 may output the clipped picture and may mark the picture as "not needed to be output". The video decoder 30 may check the buffer of the DPB that stores the clipped and output pictures. If the picture is not included in the reference picture set, the video decoder 30 may clear the buffer and may decrement the DPB fullness by one.

Although the above techniques for DPB management are described in the context of video decoder 30, in some examples, video encoder 20 may implement similar techniques. However, video encoder 20 need not implement similar techniques in every example. In some examples, video decoder 30 may implement these techniques, and video encoder 20 may not implement these techniques.

In this manner, a video coder (e.g., video encoder 20 or video decoder 30) may decode information indicating reference pictures that belong to a reference picture set. Furthermore, a reference picture set may identify reference pictures that may potentially be used for inter-prediction of a current picture and that may potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order.

The video coder may derive the reference picture set in any manner, including the example techniques described above. The video coder may determine whether a decoded picture stored in a decoded picture buffer does not need to be output and is not identified in the reference picture set. When a decoded picture has been output and is not identified in the reference picture set, the video coder may remove the decoded picture from the decoded picture buffer. After removing the decoded picture, the video coder may decode the current picture. For example, the video coder may construct reference picture list(s) as described above and decode the current picture based on the reference picture list(s).

The previous examples describe techniques that video encoder 20 and video decoder 30 may use to derive reference picture sets, construct reference picture lists from reference picture sets when modification is not required and when modification is required, and techniques for decoded picture buffer (DPB) management. However, aspects of this disclosure should not be so limited. In some examples, the techniques described in this disclosure may relate to how video encoder 20 signals which pictures belong to a reference picture set and are long-term reference pictures (or, in other words, which pictures belong to a long-term reference picture set), and how video decoder 30 determines which pictures belong to a long-term reference picture set.

[00105] For example, Table 2 includes the num_long_term_ref_pics_pps and long_term_ref_pic_id_pps[i] syntax elements as part of a picture parameter set. However, aspects of this disclosure should not be so limited. In some other examples, a sequence parameter set (e.g., Table 1) may include the num_long_term_ref_pics_pps and long_term_ref_pic_id_pps[i] syntax elements. In examples where a sequence parameter set includes these syntax elements, this disclosure may refer to syntax elements such as num_long_term_ref_pics_sps and long_term_ref_pic_id_sps[i] to avoid confusion. For illustrative purposes, the techniques are described using an example where a sequence parameter set includes these syntax elements.

Similar to the definition of num_long_term_ref_pics_pps, the num_long_term_ref_pics_sps syntax element may specify the number of candidate long-term reference pictures included in the sequence parameter set. The value of num_long_term_ref_pics_sps may be in the range of 0 to 32 (inclusive). Similar to the definition of the long_term_ref_pic_id_pps[i] syntax element, the long_term_ref_pic_id_sps[i] syntax element may specify the identification information of the i-th long-term reference picture included in the sequence parameter set.

In some examples, the long_term_ref_pic_id_sps[i] syntax element may indicate a candidate long-term reference picture that belongs to the reference picture set of the current picture. A candidate long-term reference picture is one or more of candidate long-term reference pictures that may be long-term reference pictures that video decoder 30 may use for inter-prediction of the current picture or one or more pictures that follow the current picture in decoding order. In other words, a candidate long-term reference picture may indicate a picture that is a long-term reference picture and may be used for inter-prediction of the current picture and one or more pictures that follow the current picture in decoding order. In some examples, the long_term_ref_pic_id_sps[i] syntax element may include a POC value for the candidate long-term reference picture.

However, not all candidate long-term reference pictures are used for inter-frame prediction. For example, not all candidate long-term reference pictures belong to the reference picture set of the current picture. Instead, zero or more of the candidate long-term reference pictures belong to the reference picture set.

In the techniques described in this disclosure, video encoder 20 may signal a long_term_ref_pic_id syntax element in a parameter set (e.g., a long_term_ref_pic_id_sps syntax element in a sequence parameter set, or a long_term_ref_pic_id_pps syntax element in a picture parameter set). Video decoder 30 may receive the long_term_ref_pic_id syntax element and identify candidate long-term reference pictures. According to the techniques described in this disclosure, video decoder 30 may further determine which of the candidate long-term reference pictures belong to a reference picture set. For example, video decoder 30 may be configured to perform this determination based on additional syntax elements signaled by video encoder 20 in the coded bitstream.

As indicated in Table 4, video encoder 20 may signal a long_term_ref_pic_set() syntax structure in the slice header of the current picture. Table 5 describes the long_term_ref_pic_set() syntax structure. For example, the long_term_ref_pic_set() syntax structure may include the num_long_term_pps_curr and num_long_term_pps_foll syntax elements. Furthermore, it should be noted that although the num_long_term_pps_curr and num_long_term_pps_foll syntax elements are defined as the number of long-term reference pictures included in the picture parameter set, in instances where the candidate long-term reference pictures are included in the sequence parameter set, these syntax elements may define the number of candidate long-term reference pictures included in the sequence parameter set. For example, to avoid confusion, the num_long_term_pps_curr syntax element may be referred to as the num_long_term_sps_curr syntax element, and the num_long_term_pps_foll syntax element may be referred to as the num_long_term_sps_curr syntax element.

Similar to the num_long_term_pps_curr syntax element, the num_long_term_sps_curr syntax element may define the number of all long-term reference pictures that meet the following conditions: identification information is included in the referenced sequence parameter set as candidate long-term reference pictures that can be used for inter prediction of the current picture and one or more pictures that follow the current picture in decoding order. Similar to the num_long_term_pps_foll syntax element, the num_long_term_sps_foll syntax element may define the number of all long-term reference pictures that meet the following conditions: identification information is included in the sequence parameter set as candidate long-term reference pictures that are not used for inter prediction of the current picture and can be used for inter prediction of one or more pictures that follow the current picture in decoding order.

Furthermore, the long_term_ref_pic_set() syntax structure signaled in the slice header may include the long_term_ref_pic_set_idx_pps[i] syntax element. Furthermore, in instances where candidate long-term reference pictures are signaled in a sequence parameter set, the long_term_ref_pic_set_idx_pps[i] syntax element may be considered a long_term_ref_pic_set_idx_sps[i] syntax element. Similar to the long_term_ref_pic_set_idx_pps[i] syntax element, the long_term_ref_pic_set_idx_sps[i] syntax element may define the index into the list of candidate long-term reference picture identification information contained in the referenced sequence parameter set for the i-th long-term reference picture inherited from the reference picture parameter set to the reference picture set of the current picture. In other words, the long_term_ref_pic_set_idx_sps[i] syntax element may identify the index into the list of candidate long-term reference pictures in the sequence parameter set. Based on the index, video decoder 30 may identify a long-term reference picture among the candidate long-term reference pictures, and may determine that the identified long-term reference picture belongs to a reference picture set of the current picture.

For example, video decoder 30 may implement the following pseudo-code, similar to the pseudo-code in Table 5, to determine which of the candidate long-term reference pictures belong to the reference picture set for the current picture.

for(i=0;i<num_long_term_sps_curr+num_long_term_sps_foll;i++)

long_term_ref_pic_set_idx_sps[i]

In this manner, video decoder 30 may decode syntax elements indicating candidate long-term reference pictures identified in the parameter set. For example, if the parameter set is a sequence parameter set, video decoder 30 may decode the long_term_ref_pic_id_sps[i] syntax element indicating candidate long-term reference pictures identified in the sequence parameter set. If the parameter set is a picture parameter set, video decoder 30 may decode the long_term_ref_pic_id_pps[i] syntax element indicating candidate long-term reference pictures identified in the picture parameter set.

Video decoder 30 may also decode syntax elements that indicate which candidate long-term reference pictures identified in the parameter set belong to the reference picture set of the current picture. For example, if the parameter set is a sequence parameter set, video decoder 30 may decode the num_long_term_sps_curr, num_long_term_sps_foll, and long_term_ref_pic_set_idx_sps[i] syntax elements, and if the parameter set is a picture parameter set, video decoder 30 may decode the num_long_term_pps_curr, num_long_term_pps_foll, and long_term_ref_pic_set_idx_pps[i] syntax elements. In either example, video decoder 30 may decode these syntax elements from the slice header of the current picture.

According to the techniques described in this disclosure, the long_term_ref_pic_id_sps[i] and long_term_ref_pic_id_pps[i] syntax elements may be considered as lists of picture sequence number (POC) values for candidate long-term reference pictures belonging to a reference picture set and may be coded (i.e., encoded or decoded) as part of a parameter set (e.g., a picture parameter set and a sequence parameter set). The long_term_ref_pic_set_idx_sps[i] or long_term_ref_pic_set_idx_pps[i] syntax elements may be considered as providing an index value into a list of POC values for candidate long-term reference pictures (e.g., an index into long_term_ref_pic_id_sps[i] or long_term_ref_pic_id_pps[i]). In some examples, the long_term_ref_pic_set_idx_sps[i] or long_term_ref_pic_set_idx_pps[i] syntax elements may be coded as part of a slice header for the current picture.

The foregoing describes one manner in which video encoder 20 and video decoder 30, respectively, may encode or decode syntax elements for indicating which pictures belong to the long-term reference picture set for a current picture. Based on the syntax elements, video decoder 30 and video encoder 20 may determine which pictures belong to the long-term reference picture set for the current picture. After video decoder 30 and video encoder 20 determine which pictures belong to the long-term reference picture set, video encoder 20 and video decoder 30 may construct at least one reference picture subset from a plurality of reference picture subsets and derive the reference picture set in the manner described above. For example, based on the determination of which pictures belong to the long-term reference picture set, video encoder 20 and video decoder 30 may construct a RefPicSetLtCurr reference picture subset that video encoder 20 and video decoder 30 use to derive the reference picture set.

In some examples, there may be pictures that belong to the long-term reference picture set that are not included in the candidate long-term reference pictures. Therefore, there may be additional ways by which video encoder 20 and video decoder 30 can determine which pictures belong to the long-term reference picture set for the current picture.

For example, as indicated in Table 5, the long_term_ref_pic_set() syntax structure of the slice header includes a long_term_ref_pic_id_delta_add[i] syntax element. This syntax element may specify the long-term reference picture identification (e.g., POC value) of the i-th long-term reference picture that is not inherited from the reference picture parameter set but is included in the reference picture set of the current picture. Furthermore, in the example of identifying a candidate long-term reference picture in a sequence parameter set, the long_term_ref_pic_id_delta_add[i] syntax element may specify the long-term reference picture identification of the i-th long-term reference picture that is not inherited from the sequence parameter set but is included in the reference picture set of the current picture.

In other words, the long_term_ref_pic_id_pps[i] or long_term_ref_pic_id_sps[i] syntax element may identify candidate long-term reference pictures, but may not necessarily identify all long-term reference pictures in the reference picture set of the current picture. For example, there may be long-term reference pictures that are not included in the list of candidate long-term reference pictures to be used for inter-frame prediction of the current picture and one or more pictures that follow the current picture in decoding order. For these long-term reference pictures, video encoder 20 and video decoder 30 may encode or decode, respectively, identification information that identifies the long-term reference pictures that belong to the reference picture set of the current picture.

[0066] For example, as indicated in Table 5, video encoder 20 and video decoder 30 may encode or decode the num_long_term_add_curr and num_long_term_add_foll syntax elements, respectively. The num_long_term_add_curr syntax element may define the number of all long-term reference pictures for which identifying information is not included in the referenced picture parameter set or sequence parameter set (as applicable), and which may be used for inter prediction of the current picture and one or more pictures that follow the current picture in decoding order. The num_long_term_add_foll syntax element may define the number of all long-term reference pictures for which identifying information is not included in the referenced picture parameter set or sequence parameter set (as applicable), and which may be used for inter prediction of the current picture and one or more pictures that follow the current picture in decoding order.

Video decoder 30 may implement the following pseudo-code to determine which long-term reference pictures belong to a reference picture set.In this example, the long-term reference picture may not be included in the candidate long-term reference pictures.

for(i=0;i<num_long_term_add_curr+num_long_term_add_foll;i++)

long_term_ref_pic_id_delta_add[i]

As described above, the techniques described in this disclosure may be performed according to the HEVC standard. Below is a brief description of the HEVC standard to aid understanding. Furthermore, although the techniques are described in the context of the HEVC standard, the techniques may be extended to other standards (including proprietary standards).

The JCT-VC is working on the development of the HEVC standard. The HEVC standardization effort is based on an evolving model of a video coding device, known as the HEVC Test Model (HM). The HM assumes several additional capabilities of video coding devices relative to existing devices based on, for example, ITU-T H.264/AVC. For example, while H.264 provides nine intra-frame prediction coding modes, the HM can provide up to 33 intra-frame prediction coding modes.

In general, the HM's working model describes that a video frame or image can be divided into a sequence of treeblocks or largest coding units (LCUs) that include both luma and chroma samples. A treeblock has a purpose similar to that of a macroblock in the H.264 standard. A slice includes a number of treeblocks that are consecutive in decoding order. A video frame or image can be partitioned into one or more slices. Each treeblock can be split into multiple coding units (CUs) according to a quadtree. For example, a treeblock (as the root node of the quadtree) can be split into four child nodes, and each child node can be a parent node and split into another four child nodes. The final unsplit child node (as a leaf node of the quadtree) includes a decoding node (i.e., a decoded video block). Syntax data associated with the decoded bitstream can define the maximum number of times a treeblock can be split, and can also define the minimum size of a decoding node. In some examples, a treeblock can be referred to as an LCU.

A CU includes a coding node and multiple prediction units (PUs) and transform units (TUs) associated with the coding node. The size of the CU corresponds to the size of the coding node and must be square in shape. The size of a CU can range from 8×8 pixels to the size of a treeblock with a maximum of 64×64 pixels or larger. Each CU may contain one or more PUs and one or more TUs. The syntax data associated with a CU may describe, for example, the partitioning of the CU into one or more PUs. The partitioning mode may differ depending on whether the CU is skipped or directly coded, intra-prediction mode coded, or inter-prediction mode coded. The shape of a PU may be partitioned into non-square shapes. The syntax data associated with a CU may also describe, for example, the partitioning of the CU into one or more TUs according to a quadtree. The shape of a TU may be square or non-square.

The HEVC standard allows for transforms based on TUs, which can be different for different CUs. The TU size is typically set based on the size of the PUs within a given CU defined for a partitioned LCU, but this may not always be the case. A TU is typically the same size as a PU, or smaller than a PU. In some examples, a quadtree structure called a "residual quadtree" (RQT) may be used to subdivide the residual samples corresponding to a CU into smaller units. The leaf nodes of the RQT may be referred to as transform units (TUs). Pixel difference values associated with a TU may be transformed to produce transform coefficients that may be quantized.

In general, a PU includes data related to the prediction process. For example, when the PU is encoded in intra mode, the PU may include data describing the intra prediction mode of the PU. As another example, when the PU is encoded in inter mode, the PU may include data defining a motion vector for the PU. The data defining the motion vector of the PU may describe, for example, the horizontal component of the motion vector, the vertical component of the motion vector, the resolution of the motion vector (e.g., quarter-pixel precision or eighth-pixel precision), the reference picture to which the motion vector points, and/or the reference picture list used for the motion vector (e.g., List 0, List 1, or List C).

In general, TUs are used for the transform and quantization processes. A given CU having one or more PUs may also include one or more transform units (TUs). After prediction, video encoder 20 may calculate residual values corresponding to the PUs. The residual values include pixel difference values that can be transformed into transform coefficients that are quantized and scanned using the TUs to produce serialized transform coefficients for entropy coding. This disclosure generally uses the term "video block" to refer to a coding node of a CU. In some specific cases, this disclosure may also use the term "video block" to refer to a treeblock (i.e., LCU or CU) that includes a coding node and multiple PUs and TUs.

A video sequence typically includes a series of video frames or pictures. A group of pictures (GOP) typically includes one or more of a series of video pictures. A GOP may include syntax data describing the number of pictures included in the GOP in a header of the GOP, a header of one or more of the pictures, or elsewhere. Each slice of a picture may include slice syntax data describing the encoding mode of the respective slice. Video encoder 20 typically operates on video blocks within individual video slices to encode the video data. A video block may correspond to a coding node within a CU. A video block may have a fixed or varying size and may differ in size according to a specified coding standard.

As an example, the HM supports prediction with various PU sizes. Assuming a particular CU is 2N×2N in size, the HM supports intra prediction with PU sizes of 2N×2N or N×N, and inter prediction with symmetric PU sizes of 2N×2N, 2N×N, N×2N, or N×N. The HM also supports asymmetric partitioning for inter prediction with PU sizes of 2N×nU, 2N×nD, nL×2N, and nR×2N. In an asymmetric partitioning, one direction of the CU is unpartitioned, while the other direction is partitioned into 25% and 75%. The portion of the CU corresponding to the 25% partition is indicated by an "n" followed by an indication of "up," "down," "left," or "right." Thus, for example, "2N×nU" refers to a 2N×2N CU partitioned horizontally with a 2N×0.5N PU at the top and a 2N×1.5N PU at the bottom.

In this disclosure, "NxN" and "N by N" may be used interchangeably to refer to the pixel dimensions of a video block in terms of vertical and horizontal dimensions, e.g., 16x16 pixels or 16 by 16 pixels. Generally, a 16x16 block will have 16 pixels in the vertical direction (y=16) and 16 pixels in the horizontal direction (x=16). Similarly, an NxN block generally has N pixels in the vertical direction and N pixels in the horizontal direction, where N represents a non-negative integer value. The pixels in a block may be arranged in columns and rows. Furthermore, a block need not necessarily have the same number of pixels in the horizontal direction as in the vertical direction. For example, a block may include NxM pixels, where M is not necessarily equal to N.

After performing intra-predictive or inter-predictive coding using the PUs of a CU, video encoder 20 may calculate residual data for the TUs of the CU. A PU may include pixel data in the spatial domain (also referred to as the pixel domain), and a TU may include coefficients in the transform domain after applying a transform, such as a discrete cosine transform (DCT), an integer transform, a wavelet transform, or a conceptually similar transform, to the residual video data. The residual data may correspond to pixel differences between pixels of an unencoded image and prediction values corresponding to the PU. Video encoder 20 may form TUs including the residual data for the CU, and then transform the TUs to produce transform coefficients for the CU.

After applying any transforms to generate transform coefficients, video encoder 20 may perform quantization of the transform coefficients. Quantization generally refers to the process of quantizing transform coefficients to potentially reduce the amount of data used to represent the coefficients, thereby providing further compression. The quantization process may reduce the bit depth associated with some or all of the coefficients. For example, an n-bit value may be truncated to an m-bit value during quantization, where n is greater than m.

In some examples, video encoder 20 may utilize a predefined scan order to scan the quantized transform coefficients to produce a serialized vector that can be entropy encoded. In other examples, video encoder 20 may perform adaptive scanning. After scanning the quantized transform coefficients to form a one-dimensional vector, video encoder 20 may entropy encode the one-dimensional vector, for example, according to context-adaptive variable length coding (CAVLC), context-adaptive binary arithmetic coding (CABAC), syntax-based context-adaptive binary arithmetic coding (SBAC), probability interval partitioning entropy (PIPE) coding, or another entropy encoding method. Video encoder 20 may also entropy encode syntax elements associated with the encoded video data for use by video decoder 30 in decoding the video data.

To perform CABAC, video encoder 20 may assign context within a context model to a symbol to be transmitted. The context may relate to, for example, whether neighboring values of the symbol are non-zero. To perform CAVLC, video encoder 20 may select a variable length code for the symbol to be transmitted. Codewords in VLC may be constructed in such a way that relatively shorter codes correspond to more likely symbols and longer codes correspond to less likely symbols. In this way, using VLC may achieve bit savings (compared to, for example, using equal-length codewords for each symbol to be transmitted). Probability determinations may be made based on the context assigned to the symbol.

FIG2 is a conceptual diagram illustrating an example video sequence 33 comprising a plurality of pictures that are encoded and transmitted. In some cases, video sequence 33 may be referred to as a group of pictures (GOP). As illustrated, video sequence 33 includes pictures 35A, 36A, 38A, 35B, 36B, 38B, and 35C, and a final picture 39 in display order. Picture 34 is the final picture in display order of a sequence that occurs before sequence 33. FIG2 generally represents an exemplary prediction structure for a video sequence and is intended only to illustrate picture references used to predict video blocks of different slice or picture types (e.g., P pictures or slices, or B pictures or slices). Actual video sequences may contain more or fewer video pictures of different picture types and in different display orders. Video sequence 33 may include a greater or lesser number of pictures than illustrated in FIG2 , and the pictures illustrated in video sequence 33 are illustrated for purposes of understanding and example only.

For block-based video coding, each of the video pictures included in sequence 33 may be partitioned into video blocks, such as coding units (CUs) or prediction units (PUs). For example, each CU of a video picture may include one or more PUs. Video blocks in intra-coded (I) pictures are predicted using spatial prediction with respect to neighboring blocks in the same picture. Video blocks in inter-coded (P or B) pictures may use spatial prediction with respect to neighboring blocks in the same picture or temporal prediction with respect to other reference pictures.

Video blocks in a B-picture may be predicted using bidirectional prediction to calculate two motion vectors from two different reference picture lists (e.g., reference picture lists 0 and 1, referred to as List 0 and List 1). In some cases, video blocks in a B-picture may be predicted using unidirectional prediction from one of the two different reference picture lists (e.g., unidirectional B-coded). Video blocks in a P-picture may be predicted using unidirectional prediction to calculate a single motion vector from a single reference picture list. According to the emerging HEVC standard, video blocks may be encoded using unidirectional prediction to calculate a single motion vector from one of the two reference picture lists, or using bidirectional prediction to calculate two motion vectors from both reference picture lists. For example, the two reference picture lists may contain past reference pictures, future reference pictures, or both past and future reference pictures in display or output order, and always past reference pictures in decoding order.

In the example of FIG2 , final image 39 is designated for intra-mode coding as an I picture. In other examples, final image 39 may be coded in inter-mode coding (e.g., as a P picture) with reference to final image 34 of the previous sequence (which may be an I picture). Video pictures 35A-35C (collectively, "video pictures 35") are designated for coding as B pictures using bidirectional prediction with reference to past and future pictures. In the illustrated example, picture 35A is encoded as a B picture with reference to final picture 34 and picture 36A, as indicated by the arrows from pictures 34 and 36A to video picture 35A. Pictures 35B and 35C are similarly encoded.

Video pictures 36A-36B (collectively, "video pictures 36") may be designated for coding as pictures using unidirectional prediction with reference to past pictures. In the illustrated example, picture 36A is encoded as a B picture with reference to final picture 34, as indicated by the arrow from picture 34 to video picture 36A. Picture 36B is similarly encoded.

Video pictures 38A-38B (collectively, "video pictures 38") may be designated for coding using bidirectional prediction with reference to the same past picture. In other examples, video pictures 38 may be encoded using bidirectional prediction with reference to substantially similar past pictures included in a reference picture list. In the illustrated example, picture 38A is encoded with two references to picture 36A, as indicated by the two arrows from picture 36A to video picture 38A. Picture 38B is similarly encoded.

According to the techniques described in this disclosure, video encoder 20 may signal a reference picture set for each of the pictures in sequence 33. For example, for picture 35A, this reference picture set may identify all reference pictures that can be used to inter-predict picture 35A, as well as all reference pictures that may potentially be used to inter-predict pictures that follow picture 35A in decoding order. For example, the reference picture set for picture 35A may include POC values for pictures 34 and 36A, as well as POC values for additional reference pictures (e.g., reference pictures that may potentially be used to inter-predict pictures that follow picture 35A in decoding order). In this example, the pictures that follow picture 35A may be those pictures that follow picture 35A in decoding order and that are within video sequence 33.

Video decoder 30 may then derive a reference picture set for image 35A in the manner described above. For example, video decoder 30 may determine POC values for reference pictures belonging to the reference picture set, as described above. Video decoder 30 may further construct at least four or at least five reference picture subsets, and in some examples, up to the six reference picture subsets described above. Video decoder 30 may arrange the six reference picture subsets in a particular order to derive the reference picture set for image 35A.

The video decoder 30 may further construct the initial reference picture list in the manner described above, without reordering the pictures to be included in the initial reference picture list. When reference picture list modification is disabled, the video decoder 30 may set the final reference picture list equal to the initial reference picture list. Furthermore, the video decoder 30 may construct the reference picture list in such a way that no incomplete entries exist in the reference picture list. For example, the video decoder 30 may repeatedly list reference pictures from the reference picture subset until the number of entries in the reference picture list equals the maximum number of allowable entries in the reference picture list. In some examples, the video decoder 30 may modify the initial reference picture list in the manner described above (e.g., based on reference pictures in at least one of the constructed reference picture subsets). Furthermore, in some examples, the video decoder 30 may utilize example techniques described in this disclosure to remove decoded pictures from the DPB of the video decoder 30, for example, to remove decoded pictures that were not identified in the reference picture set of the current picture to be decoded and that do not need to be output. Moreover, in some examples, video decoder 30 may determine which long-term reference pictures belong to a reference picture set in the manner described above, where identification information of a candidate long-term reference picture list may be included in a parameter set.

FIG3 is a block diagram illustrating an example video encoder 20 that may implement the techniques described in this disclosure. Video encoder 20 may perform intra- and inter-coding of video blocks within a video slice. Intra-coding relies on spatial prediction to reduce or remove spatial redundancy in video within a given video frame or picture. Inter-coding relies on temporal prediction to reduce or remove temporal redundancy in video within adjacent frames or pictures of a video sequence. Intra-mode (I-mode) may refer to any of several spatially-based compression modes. Inter-modes, such as uni-directional prediction (P-mode) or bi-directional prediction (B-mode), may refer to any of several temporally-based compression modes.

In the example of FIG3 , video encoder 20 includes a partitioning unit 35, a prediction module 41, a decoded picture buffer (DPB) 64, a summer 50, a transform module 52, a quantization unit 54, and an entropy encoding unit 56. Prediction module 41 includes a motion estimation unit 42, a motion compensation unit 44, and an intra-prediction module 46. For video block reconstruction, video encoder 20 also includes an inverse quantization unit 58, an inverse transform module 60, and a summer 62. A deblocking filter (not shown in FIG3 ) may also be included to filter block boundaries to remove blocking artifacts from the reconstructed video. If necessary, the deblocking filter typically filters the output of summer 62. In addition to the deblocking filter, additional loop filters (in-loop or post-loop) may also be used.

As shown in FIG3 , video encoder 20 receives video data, and partitioning unit 35 partitions the data into video blocks. This partitioning may also include partitioning into slices, tiles, or other larger units, as well as video block partitioning according to a quadtree structure of LCUs and CUs, for example. Video encoder 20 generally illustrates components that encode video blocks within a video slice to be encoded. A slice may be divided into multiple video blocks (and potentially into sets of video blocks referred to as tiles). Prediction module 41 may select one of multiple possible coding modes for the current video block, such as one of multiple intra-coding modes or one of multiple inter-coding modes, based on error results (e.g., coding rate and distortion level). Prediction module 41 may provide the resulting intra-coded or inter-coded block to summer 50 to generate residual block data, and provide the resulting intra-coded or inter-coded block to summer 62 to reconstruct the encoded block for use as a reference picture.

Intra-prediction module 46 within prediction module 41 may perform intra-predictive coding of the current video block relative to one or more neighboring blocks in the same picture or slice as the current block to be coded to provide spatial compression. Motion estimation unit 42 and motion compensation unit 44 within prediction module 41 may perform inter-predictive coding of the current video block relative to one or more predictive blocks in one or more reference pictures to provide temporal compression.

Motion estimation unit 42 may be configured to determine an inter-prediction mode for a video slice according to a predetermined pattern for a video sequence. The predetermined pattern may specify whether a video slice in the sequence is a P slice or a B slice. Motion estimation unit 42 and motion compensation unit 44 may be highly integrated but are illustrated separately for conceptual purposes. Motion estimation, performed by motion estimation unit 42, is the process of generating motion vectors, which estimate motion for video blocks. A motion vector may, for example, indicate the displacement of a PU of a video block within a current video picture relative to a predictive block within a reference picture.

A predictive block is a block that is found to closely match a PU of the video block to be coded in terms of pixel difference, which may be determined by sum of absolute difference (SAD), sum of squared difference (SSD), or other different metrics. In some examples, video encoder 20 may calculate values for sub-integer pixel positions of a reference picture stored in decoded picture buffer 64. For example, video encoder 20 may interpolate values for quarter-pixel positions, eighth-pixel positions, or other fractional pixel positions of the reference picture. Thus, motion estimation unit 42 may perform motion searches relative to full-pixel positions and fractional pixel positions and output motion vectors with fractional pixel precision.

Motion estimation unit 42 calculates a motion vector for a PU of a video block in an inter-coded slice by comparing the position of the PU to the position of a predictive block of a reference picture. The reference picture may be selected from a first reference picture list (List 0) or a second reference picture list (List 1), each of which identifies one or more reference pictures stored in decoded picture buffer 64. Motion estimation unit 42 sends the calculated motion vector to entropy encoding unit 56 and motion compensation unit 44.

Motion compensation performed by motion compensation unit 44 may involve retrieving or generating a predictive block based on the motion vector determined by motion estimation, possibly performing interpolation with sub-pixel precision. Upon receiving the motion vector for the PU of the current video block, motion compensation unit 44 may locate the predictive block pointed to by the motion vector in one of the reference picture lists. Video encoder 20 forms a residual video block by subtracting the pixel values of the predictive block from the pixel values of the current video block being coded, forming pixel difference values. The pixel difference values form residual data for the block and may include both luma and chroma difference components. Summer 50 represents one or more components that perform this subtraction operation. Motion compensation unit 44 may also generate syntax elements associated with the video block and video slice for use by video decoder 30 in decoding the video block of the video slice.

Intra-prediction module 46 may intra-predict the current block, as an alternative to the inter-prediction performed by motion estimation unit 42 and motion compensation unit 44 (as described above). In particular, intra-prediction module 46 may determine an intra-prediction mode to use to encode the current block. In some examples, intra-prediction module 46 may encode the current block using various intra-prediction modes (e.g., during separate encoding passes), and intra-prediction module 46 (or, in some examples, mode selection unit 40) may select an appropriate intra-prediction mode to use from the tested modes. For example, intra-prediction module 46 may calculate rate-distortion values using a rate-distortion analysis of the various tested intra-prediction modes and select the intra-prediction mode with the best rate-distortion characteristics among the tested modes. The rate-distortion analysis generally determines the amount of distortion (or error) between the encoded block and the original, unencoded block that was encoded to produce the encoded block, as well as the bit rate (i.e., the number of bits) used to produce the encoded block. Intra-prediction module 46 may calculate ratios from the distortions and rates for the various encoded blocks to determine which intra-prediction mode exhibits the best rate-distortion value for the block.

After selecting an intra-prediction mode for a block, intra-prediction module 46 may provide information indicating the selected intra-prediction mode for the block to entropy encoding unit 56. Entropy encoding unit 56 may encode the information indicating the selected intra-prediction mode according to the techniques of this disclosure. Video encoder 20 may include definitions of coding contexts for the various blocks and indications of the most probable intra-prediction mode to be used for each of the contexts, an intra-prediction mode index table, and a modified intra-prediction mode index table in the transmitted bitstream configuration data, which may include multiple intra-prediction mode index tables and multiple modified intra-prediction mode index tables (also referred to as codeword mapping tables).

After prediction module 41 generates a predictive block for the current video block via inter-prediction or intra-prediction, video encoder 20 forms a residual video block by subtracting the predictive block from the current video block. The residual video data in the residual block may be included in one or more TUs and applied to transform module 52. Transform module 52 transforms the residual video data into residual transform coefficients using a transform, such as a discrete cosine transform (DCT) or a conceptually similar transform. Transform module 52 may convert the residual video data from the pixel domain to a transform domain, such as the frequency domain.

Transform module 52 may send the resulting transform coefficients to quantization unit 54. Quantization unit 54 quantizes the transform coefficients to further reduce the bit rate. The quantization process may reduce the bit depth associated with some or all of the coefficients. The degree of quantization may be modified by adjusting a quantization parameter. In some examples, quantization unit 54 may then perform a scan of the matrix containing the quantized transform coefficients. Alternatively, entropy encoding unit 56 may perform the scan.

After quantization, entropy encoding unit 56 entropy encodes the quantized transform coefficients. For example, entropy encoding unit 56 may perform context adaptive variable length coding (CAVLC), context adaptive binary arithmetic coding (CABAC), syntax-based context adaptive binary arithmetic coding (SBAC), probability interval partitioning entropy (PIPE) coding, or another entropy encoding method or technique. Following entropy encoding by entropy encoding unit 56, the encoded bitstream may be transmitted to video decoder 30 or archived for later transmission or retrieval by video decoder 30. Entropy encoding unit 56 may also entropy encode the motion vectors and other syntax elements for the current video slice being coded.

Inverse quantization unit 58 and inverse transform module 60 apply inverse quantization and inverse transform, respectively, to reconstruct the residual block in the pixel domain for later use as a reference block for a reference picture. Motion compensation unit 44 may calculate a reference block by adding the residual block to a predictive block of one of the reference pictures in one of the reference picture lists. Motion compensation unit 44 may also apply one or more interpolation filters to the reconstructed residual block to calculate sub-integer pixel values for use in motion estimation. Summer 62 adds the reconstructed residual block to the motion compensated prediction block produced by motion compensation unit 44 to produce a reference block for storage in decoded picture buffer 64. The reference block may be used by motion estimation unit 42 and motion compensation unit 44 as a reference block for inter-frame prediction of blocks in subsequent video frames or pictures.

According to this disclosure, prediction module 41 represents one example unit for performing the example functions described above. For example, prediction module 41 may determine which reference pictures belong to a reference picture set and cause video encoder 20 to decode information indicating the reference pictures belonging to the reference picture set. Furthermore, during a reconstruction process (e.g., a process used to reconstruct pictures for use as reference pictures and stored in decoded picture buffer 64), prediction module 41 may construct multiple reference picture subsets, each of which identifies one or more of the reference pictures. Prediction module 41 may also derive a reference picture set from the constructed multiple reference picture subsets. Furthermore, prediction module 41 may implement any one or more of the example pseudo-code sets described above to implement one or more example techniques described in this disclosure.

In some examples, prediction module 41 may construct an initial reference picture list in the manner described above. In some examples, there is no need to reorder the pictures to be included in the initial reference picture list. Furthermore, prediction module 41 may construct the reference picture list in such a way that no incomplete entries exist in the reference picture list. In some examples, prediction module 41 may also modify the initial reference picture list in the manner described above to construct a modified reference picture list. Furthermore, in some examples, prediction module 41 may implement the removal of decoded pictures from DPB 64 in the manner described above. Furthermore, in some examples, prediction module 41 may be configured to determine which long-term reference pictures belong to the reference picture set of the current picture in the manner described above.

In other examples, units other than prediction module 41 may implement the examples described above. In some other examples, prediction module 41, in conjunction with one or more other units of video encoder 20, may implement the examples described above. In still other examples, a processor or unit (not shown in FIG. 3 ) of video encoder 20, alone or in conjunction with other units of video encoder 20, may implement the examples described above.

FIG4 is a block diagram illustrating an example video decoder 30 that may implement the techniques described in this disclosure. In the example of FIG4, video decoder 30 includes an entropy decoding unit 80, a prediction module 81, an inverse quantization unit 86, an inverse transform unit 88, a summer 90, and a decoded picture buffer (DPB) 92. Prediction module 81 includes a motion compensation unit 82 and an intra prediction module 84. Video decoder 30 may, in some examples, perform a decoding pass that is generally reciprocal to the encoding pass described with respect to video encoder 20 (FIG. 3).

During the decoding process, video decoder 30 receives an encoded video bitstream representing video blocks of an encoded video slice and associated syntax elements from video encoder 20. Entropy decoding unit 80 of video decoder 30 entropy decodes the bitstream to generate quantized coefficients, motion vectors, and other syntax elements. Entropy decoding unit 80 forwards the motion vectors and other syntax elements to prediction module 81. Video decoder 30 may receive syntax elements at the video slice level and/or at the video block level.

When the video slice is coded as an intra-coded (I) slice, intra-prediction module 84 of prediction module 81 may generate prediction data for the video block of the current video slice based on the signaled intra-prediction mode and data from previously decoded blocks of the current picture. When the video picture is coded as an inter-coded (i.e., B or P) slice, motion compensation unit 82 of prediction module 81 generates a predictive block for the video block of the current video slice based on the motion vector and other syntax elements received from entropy decoding unit 80. The predictive block may be generated from one of the reference pictures in one of the reference picture lists. Video decoder 30 may construct the reference frame lists (List 0 and List 1) based on the reference pictures stored in decoded picture buffer 92 using default construction techniques. In some examples, video decoder 30 may construct List 0 and List 1 from reference pictures identified in the derived reference picture set.

Motion compensation unit 82 determines prediction information for a video block of the current video slice by analyzing motion vectors and other syntax elements, and uses the prediction information to produce a predictive block for the current video block being decoded. For example, motion compensation unit 82 uses some of the received syntax elements to determine: a prediction mode (e.g., intra-prediction or inter-prediction) used to code the video block of the video slice, an inter-prediction slice type (e.g., B-slice or P-slice), construction information for one or more of the reference picture lists for the slice, a motion vector for each inter-coded video block of the slice, an inter-prediction state for each inter-coded video block of the slice, and other information used to decode the video blocks in the current video slice.

Motion compensation unit 82 may also perform interpolation based on interpolation filters. Motion compensation unit 82 may use interpolation filters as used by video encoder 20 during encoding of video blocks to calculate interpolated values for sub-integer pixels of reference blocks. In this case, motion compensation unit 82 may determine the interpolation filters used by video encoder 20 from received syntax elements and use the interpolation filters to produce predictive blocks.

Inverse quantization unit 86 inverse quantizes (i.e., dequantizes) the quantized transform coefficients provided in the bitstream and decoded by entropy decoding unit 80. The inverse quantization process may include using the quantization parameter for each video block in the video slice calculated by video encoder 20 to determine the degree of quantization and, likewise, the degree of inverse quantization that should be applied. Inverse transform module 88 applies an inverse transform (e.g., an inverse DCT, an inverse integer transform, or a conceptually similar inverse transform process) to the transform coefficients to produce residual blocks in the pixel domain.

After prediction module 81 generates a predictive block for the current video block based on inter-frame prediction or intra-frame prediction, video decoder 30 forms a decoded video block by summing the residual block from inverse transform module 88 with the corresponding predictive block generated by prediction module 81. Summer 90 represents one or more components that perform this summing operation. If desired, a deblocking filter may also be applied to filter the decoded block to remove blocking artifacts. Other loop filters (in the decoding loop or after the decoding loop) may also be used to smooth pixel transitions or otherwise improve video quality. The decoded video block in a given picture is then stored in decoded picture buffer 92, which stores reference pictures used for subsequent motion compensation. Decoded picture buffer 92 also stores the decoded video for later presentation on a display device (e.g., display device 32 of FIG. 1 ).

According to this disclosure, prediction module 81 represents one example unit for performing the example functions described above. For example, prediction module 81 may determine which reference images belong to a reference image set. Furthermore, prediction module 81 may construct multiple reference image subsets, each of which identifies one or more of the reference images. Prediction module 81 may also derive a reference image set from the constructed multiple reference image subsets. Furthermore, prediction module 81 may implement any one or more of the example pseudo-code sets described above to implement one or more example techniques described in this disclosure.

In some examples, the prediction module 81 may construct an initial reference picture list in the manner described above. In some examples, there is no need to reorder the pictures to be included in the initial reference picture list. Furthermore, the prediction module 81 may construct the reference picture list in such a way that no incomplete entries exist in the reference picture list. In some examples, the prediction module 81 may also modify the initial reference picture list in the manner described above to construct a modified reference picture list. Furthermore, in some examples, the prediction module 81 may implement the removal of decoded pictures from the DPB 92 in the manner described above. Furthermore, in some examples, the prediction module 81 may be configured to determine which long-term reference pictures belong to the reference picture set of the current picture in the manner described above.

In other examples, units other than prediction module 81 may implement the examples described above. In some other examples, prediction module 81, in conjunction with one or more other units of video decoder 30, may implement the examples described above. In still other examples, a processor or unit (not shown in FIG. 4 ) of video decoder 30, alone or in conjunction with other units of video decoder 30, may implement the examples described above.

FIG5 is a flowchart illustrating an example operation of deriving a reference picture set. For purposes of illustration only, the method of FIG5 may be performed by a video coder corresponding to video encoder 20 or video decoder 30. For example, a video coder (e.g., video encoder 20 or video decoder 30) may code (e.g., encode or decode) information indicating reference pictures belonging to a reference picture set (94). The reference picture set may identify reference pictures that may potentially be used for inter-prediction of a current picture and for inter-prediction of one or more pictures that follow the current picture in decoding order.

For example, when video encoder 20 performs step 94, video encoder 20 may encode a value indicating an identifier of a reference picture belonging to a reference picture set. For example, video encoder 20 may signal a pic_order_cnt_lsb syntax element and a log2_max_pic_order_cnt_lsb_minus4 syntax element in the bitstream. When video decoder 30 performs step 94, video decoder 30 may determine a value of MaxPicOrderCntLsb based on the log2_max_pic_order_cnt_lsb_minus4 syntax element. Video decoder 30 may then determine an identifier (e.g., a POC value) of the reference picture belonging to the reference picture set.

The video coder may construct a plurality of reference picture subsets, each reference picture subset identifying zero or more of the reference pictures (96). For example, the video coder may construct the RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, RefPicSetStFoll1, RefPicSetLtCurr, and RefPicSetLtFoll reference picture subsets. However, aspects of this disclosure should not be so limited. In some examples, the video coder may construct five reference picture subsets, four of which are four of the RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, RefPicSetStFoll1, RefPicSetLtCurr, and RefPicSetLtFoll reference picture subsets, and the fifth may be a combination of two of the remaining six reference picture subsets (e.g., a combination of the RefPicSetFoll0 and RefPicSetFoll1 reference picture subsets).

In some examples, the video coder may construct at least two of the following four reference picture subsets. In other examples, the video coder may construct at least the following four reference picture subsets. A first reference picture subset may identify short-term reference pictures that precede a current picture in decoding order and precede the current picture in output order, and that may potentially be used for inter-prediction of the current picture and one or more pictures that follow the current picture in decoding order. A second reference picture subset may identify short-term reference pictures that precede the current picture in decoding order and follow the current picture in output order, and that may potentially be used for inter-prediction of the current picture and one or more pictures that follow the current picture in decoding order.

The third reference picture subset may identify long-term reference pictures that precede the current picture in decoding order and that can potentially be used for inter-prediction of the current picture and one or more pictures that follow the current picture in decoding order. The fourth reference picture subset may identify long-term reference pictures that precede the current picture in decoding order and that cannot be used for inter-prediction of the current picture, and that can potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order.

The video coder may derive a reference picture set from the plurality of reference picture subsets (98). For example, the video coder may sort at least two of the following in a particular order: RefPicSetStCurr0, RefPicSetStCurr1, RefPicSetStFoll0, RefPicSetStFoll1, RefPicSetLtCurr, and RefPicSetLtFoll reference picture subsets to derive the reference picture set.

In some examples, the ordering performed by the video coder may mean that the images in each of the reference image subsets can be identified sequentially within the reference image set. In these examples, the video coder can reference a reference image in the reference image set by an index value into the reference image set.

A video coder may code a current picture based on the derived reference picture set (100). It should be understood that because the video coder derives the reference picture set from the reference picture subsets, the video coder may be considered to code the current picture based on the multiple reference picture subsets. For example, the video coder may construct at least one of a first reference picture list and a second reference picture list based on the multiple reference picture subsets (e.g., from the derived reference picture set derived from the multiple reference picture subsets). The video coder may then code the current picture based on at least one of the first reference picture list and the second reference picture list.

FIG6 is a flowchart illustrating an example operation of constructing a reference picture list. For illustrative purposes only, the method of FIG6 may be performed by a video coder corresponding to video encoder 20 or video decoder 30. Similar to FIG5 , the video coder may decode information indicating a reference picture (102) and construct a plurality of reference picture subsets (104).

The video coder may then add the reference pictures from the reference picture subset to the initial reference picture list to construct the initial reference picture list (106). In some examples, both video encoder 20 and video decoder 30 may construct the initial reference picture list. For example, video encoder 20 may construct the initial reference picture list to create a reconstructed video block for storage in DPB 64. Video decoder 30 may construct the initial reference picture list as part of its decoding process and may implement a default construction technique in which video decoder 30 does not need to receive information from video encoder 20 regarding the manner in which the initial reference picture list was constructed.

In some examples, to construct an initial reference picture list, the video coder may add reference pictures from a first subset of the plurality of reference picture subsets to the initial reference picture list, followed by adding reference pictures from a second subset to the initial reference picture list, and then subsequently adding reference pictures from a third subset to the initial reference picture list. The video coder may add reference pictures from these reference picture subsets as long as the total number of reference pictures listed in the initial reference picture list is not greater than the maximum number of allowable entries in the initial reference picture list. For example, if at any time while adding reference pictures to the reference picture list, the number of entries in the initial reference picture list becomes equal to the maximum number of allowable initial reference list entries, the video coder may stop adding any additional pictures to the initial reference picture list.

The video coder may similarly construct another initial reference picture list, such as in an example where the video block of the current picture is bidirectionally predicted. In this example, to construct this another initial reference picture list, the video coder may add reference pictures from the second subset to the another initial reference picture list, followed by adding reference pictures from the first subset to the another initial reference picture list, and then subsequently adding reference pictures from the third subset to the another initial reference picture list, as long as the total number of entries in the another initial reference picture list is not greater than the allowable number of entries. In these examples, the first subset may be the RefPicSetStCurr0 reference picture subset, the second subset may be the RefPicSetStCurr1 reference picture subset, and the third subset may be the RefPicSetLtCurr reference picture subset.

In some examples, to add reference pictures identified in the RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets, a video coder may code (e.g., encode or decode) syntax elements from which the video coder may determine the number of reference pictures in each of these reference picture subsets. For example, the video coder may code the num_short_term_curr0 syntax element and the num_short_term_curr1 syntax element. The num_short_term_curr0 syntax element and the num_short_term_curr1 syntax element may indicate the number of reference pictures identified in the RefPicSetStCurr0 reference picture subset and the RefPicSetStCurr1 reference picture subset, respectively.

The video coder may also decode the num_long_term_pps_curr syntax element and the num_long_term_add_curr syntax element. The num_long_term_pps_curr syntax element may indicate the number of long-term reference pictures identified in a picture parameter set (PPS), and the num_long_term_add_curr syntax element may indicate the number of long-term reference pictures whose identification information is not included in the PPS. In this example, these long-term reference pictures may potentially be used for inter-prediction of the current picture and may potentially be used for inter-prediction of one or more pictures that follow the current picture in decoding order.

The video coder may determine the number of reference pictures in the RefPicSetLtCurr reference picture subset based on the num_long_term_pps_curr syntax element and the num_long_term_add_curr syntax element. For example, the video coder may sum the values of the num_long_term_pps_curr syntax element and the num_long_term_add_curr syntax element to determine the number of reference pictures in the RefPicSetLtCurr reference picture subset.

The video coder may code the current picture based on one or more reference picture lists (108). For example, the video coder may construct at least one of a first reference picture list and a second reference picture list based on the derived reference picture set. The video coder may then code the current picture based on at least one of the first reference picture list and the second reference picture list.

FIG7 is a flowchart illustrating another example operation of constructing a reference picture list. For purposes of illustration only, the method of FIG7 may be performed by a video coder corresponding to video encoder 20 or video decoder 30. Similar to FIG5 , the video coder may decode information indicating a reference picture (110) and construct a plurality of reference picture subsets (112). The video coder may add the reference picture to a first set of entries in the reference picture list (114). For example, the number of entries in the reference picture list may be equal to the maximum number of allowable entries, as defined by the num_ref_idx_l0_active_minus1 or num_ref_idx_l1_active_minus1 syntax elements.

In this example, the video coder may list (e.g., add) the reference pictures identified in the RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets into a first set of entries. For example, for List 0, the video coder may add the reference pictures identified in the RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets, in order, to a first set of entries for List 0. For List 1, the video coder may add the reference pictures identified in the RefPicSetStCurr1, RefPicSetStCurr0, and RefPicSetLtCurr reference picture subsets, in order, to a first set of entries for List 1.

The video coder may then determine whether the number of entries in the reference picture list is equal to the maximum number of allowable entries in the reference picture list (116). If the number of entries in the reference picture list is not less than the maximum number of allowable entries ("No" of 116), the video coder may code the current picture based on the reference picture list (118).

Otherwise, if the number of entries in the reference picture list is less than the maximum number of allowable entries ("YES" of 116), the video coder may relist (e.g., reidentify or re-add) one or more reference pictures from at least one of the reference picture subsets to entries in the reference picture list that follow the first set of entries (120). For example, the video coder may add one or more reference pictures identified in the RefPicSetStCurr0 reference picture subset to entries in List0 that follow the first set of entries in List0, or add one or more reference pictures identified in the RefPicSetStCurr1 reference picture subset to entries in List1 that follow the first set of entries in List1. In this manner, the video coder may identify at least one reference picture in the first reference picture subset in one or more entries in the reference picture list.

The video coder may then determine whether the number of entries in the reference picture list is equal to the maximum number of allowable entries in the reference picture list (122). If the number of entries in the reference picture list is not less than the maximum number of allowable entries ("No" of 122), the video coder may code the current picture based on the reference picture list (124).

Otherwise, if the number of entries in the reference picture list is less than the maximum number of allowable entries ("YES" of 122), the video coder may re-list (e.g., re-identify or re-add) one or more reference pictures from at least one of the reference picture subsets to the entries in the reference picture list that follow the first set of entries (120). For example, in this case, the video coder may re-add the additional reference pictures identified in the first reference picture subset. If the video coder has re-added all reference pictures in the first reference picture subset, the video coder may re-add one or more reference pictures from the second reference picture subset (e.g., the RefPicSetStCurr0 reference picture subset for List 0 or the RefPicSetStCurr1 reference picture subset for List 1). This process may be repeated until the number of entries in the reference picture list is no less than the maximum number of allowable entries ("NO" of 122).

In this manner, when the number of entries in the reference picture list is not equal to the maximum number of allowable entries in the reference picture list, the video coder may repeatedly re-list (e.g., re-identify or re-add) one or more reference pictures from at least one of the reference picture subsets to entries following the first set of entries in the reference picture list until the number of entries in the reference picture list is equal to the maximum number of allowable entries in the reference picture list. This may result in the video coder listing the reference pictures in the entries of the reference picture list such that each entry of the reference picture list identifies one of the reference pictures and such that at least two entries in the reference picture list identify the same reference picture of the reference pictures.

FIG8 is a flowchart illustrating an example operation of modifying an initial reference picture list. For purposes of illustration only, the method of FIG8 may be performed by a video coder corresponding to video encoder 20 or video decoder 30. Similar to FIG6 , the video coder may decode information indicating a reference picture (126) and construct a plurality of reference picture subsets (128). The plurality of reference picture subsets may include RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets. The video coder may construct an initial reference picture list (130) based on the RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr reference picture subsets in the manner described above.

The video coder may determine whether reference picture list modification is required (132). For example, the video coder may code syntax elements such as ref_pic_list_modification_flag_l0 and ref_pic_list_modification_flag_l1 that indicate whether initial list 0 or initial list 1 needs to be modified. If modification of the initial reference picture list is not required ("No" of 132), the video coder may code the current picture based on the initial reference picture list (134).

If modification is required ("YES" of 132), the video coder may identify a reference picture in at least one of the constructed reference picture subsets (136). For example, the video coder may code a modification_of_ref_pic_idc syntax element. The value of the modification_of_ref_pic_idc syntax element may indicate which reference picture subset the video coder will utilize to identify the reference picture. The video coder may also code a ref_pic_set_idx syntax element that indicates an index into the reference picture subset that the video coder will use to identify the reference picture.

The video coder may list (e.g., add or identify) the identified reference picture at the current entry in the initial reference picture list to construct a modified reference picture list (138). The current entry may initially be the entry defined by index 0 in the initial reference picture list. For each instance of the modification_of_ref_pic_idc syntax element in the coded bitstream (where the value of the modification_of_ref_pic_idc syntax element is not 3), the video coder may increment the value of the initial entry by one (e.g., define the next value of the entry by index 1). The video coder may code the current picture based on the modified reference picture list (140).

FIG9 is a flowchart illustrating an example operation of decoded picture removal. For purposes of illustration only, the method of FIG8 may be performed by a video coder corresponding to video encoder 20 or video decoder 30. Similar to FIG5 , a video coder (e.g., video encoder 20 or video decoder 30) may code (e.g., encode or decode) information indicating reference pictures of a reference picture set (142) and may derive the reference picture set from the coded (e.g., encoded or decoded) information (144).

The video coder may determine whether a decoded picture stored in a decoded picture buffer (DPB) does not need to be output (i.e., has already been output or is not desired to be output) and is not identified in a reference picture set (146). If the decoded picture needs to be output or is identified in a reference picture set ("No" of 146), the video coder may not remove the decoded picture from the DPB and may keep the decoded picture stored in the DPB (148).

When the decoded picture has been output and is not identified in the reference picture set ("YES" of 146), the video coder may remove the decoded picture from the DPB (150). After removing the decoded picture, the video coder may then decode the current picture (152).

As described above, the video coder may construct a reference picture list based on the reference picture set used to code the current picture. In some examples, the video coder may remove the decoded picture from the DPB after constructing the reference picture list. Furthermore, for the decoded picture that is output, the video coder may determine a time to output the decoded picture and may output the decoded picture based on the determined time and before coding the current picture.

In some examples, the video coder may store a coded (e.g., decoded) picture in the DPB. In some examples, the video coder may determine that the DPB is full before storing. In these examples, the video coder may select a decoded picture that satisfies the following conditions: is currently stored in the DPB, is marked as "need to be output," and has the smallest picture sequence number (POC) value of all decoded pictures stored in the DPB. The video coder may then output the selected picture. In addition, the video coder may determine that the output picture is not included in the reference picture set of the current picture. In this case, the video coder may clear the buffer within the DPB that stores the output picture and may store the coded picture in the buffer within the DPB.

10 is a flowchart illustrating an example operation of determining which long-term reference pictures belong to the reference picture set of the current picture. For purposes of illustration only, the method of FIG. 10 may be performed by a video coder corresponding to video encoder 20 or video decoder 30.

A video coder (e.g., video encoder 20 or video decoder 30) may code a syntax element indicating candidate long-term reference pictures identified in a parameter set (154). In some examples, one or more of the candidate long-term reference pictures belong to a reference picture set for a current picture. According to the techniques described in this disclosure, the parameter set may be a sequence parameter set or a picture parameter set. In some examples, to code the syntax element indicating the candidate long-term reference pictures, the video coder may code a list of POC values for the candidate long-term reference pictures in the parameter set and an index value to the list in a slice header of the current picture.

The video coder may code a syntax element indicating which candidate long-term reference pictures identified in the parameter set belong to the reference picture set of the current picture (156). For example, the video coder may code the syntax element indicating which candidate long-term reference pictures belong to the reference picture set in a slice header of the current picture.

In some examples, not all long-term reference pictures belonging to the reference picture set are included in the candidate long-term reference pictures. In these examples, the video coder may further code syntax elements indicating the long-term reference pictures belonging to the reference picture set (158). The video coder may then construct at least one of a plurality of reference picture subsets based on the indication of which candidate long-term reference pictures belong to the reference picture set for the current picture (160).

In the above examples, various examples and possible alternatives to the examples are described. Below, some additional alternatives to the examples are described, which can be considered as alternatives to one or more of the examples described above. Furthermore, these alternatives can be used in conjunction with the examples described above, or separately from them.

For example, the examples described above are alternatives to the concept of reference picture sets, alternatives to signaling reference picture sets in parameter sets, alternatives to subsets of reference picture sets, and alternatives to reference picture labeling. Additional alternatives are provided below.

For example, for the NAL unit header, in the above example, the NAL unit header syntax may include nal_ref_idc, temporal_id, and output_flag syntax elements. In an alternative example, nal_ref_idc (2 bits) is replaced with nal_ref_flag (1 bit). In this example, nal_ref_flag equal to 1 has the semantics of nal_ref_idc being greater than 0, and nal_ref_flag equal to 0 has the same semantics as nal_ref_idc being equal to 0. In an alternative example, nal_ref_idc (2 bits) is removed. The definition of a reference picture may be changed to one containing samples that can be used for inter-frame prediction during the decoding of subsequent pictures in decoding order. In an alternative example, temporal_id is not present in the NAL unit header syntax, and it is derived that its value is the same for all pictures. In an alternative example, output_flag is not present in the NAL unit header syntax, and it is derived that its value is equal to 1 for all pictures.

For picture sequence number signaling and calculation, the above-described techniques may use a method that may be similar to the signaling and calculation of picture sequence numbers for picture sequence number type 0 in AVC. Two alternative methods for signaling and calculation of picture sequence numbers, which are the same as those for picture sequence number types 1 and 2 in AVC, respectively, may be used. Any combination of two of the three methods, or a combination of all three methods, may also be used.

For image identification, the above-described techniques can utilize a picture sequence number (POC) value for image identification. In an alternative example, a temporal reference (TR) is used for image identification. One way to define TR is that when the POC is constrained so that the POC difference between any two images is proportional to the presentation time or sampling time difference, the TR is the same as the POC. In an alternative example, frame_num, which indicates the decoding order (whereas the POC indicates the output or display order), or a variable derived from frame_num (e.g., named UnWrappedFrameNum), which can be any of a 32-bit unsigned integer value, can be used for image identification. Essentially, UnWrappedFrameNum can be an expanded version of frame_num. For example, if frame_num is represented using 8 bits, the maximum value of frame_num is 255. The next value for frame_num after 255 is 0. The value of UnWrappedFrameNum continues to increase after frame_num wraps around to 0.

For signaling long-term reference pictures in parameter sets, in the above example, video encoder 20 may signal a list of absolute long-term reference picture identification information in a picture parameter set. Information regarding the length of the absolute long-term reference picture identification information in bits may be signaled, and the index to the list may be referenced in a slice header, thereby reducing signaling overhead. In an alternative example, a list of differential long-term reference picture identification information may be signaled in a sequence parameter set, and the index to the list may be referenced in a slice header, thereby reducing signaling overhead. In an alternative example, information regarding the length of the absolute long-term reference picture identification information in bits may be signaled, but the length may be considered a constant (e.g., 32). In various examples, combinations of any of the above examples apply.

Regarding the signaling of short-term reference pictures, in the above example, for the short-term reference pictures in the reference picture set of the coded picture, either all the short-term reference pictures are signaled in the referenced picture parameter set or all the short-term reference pictures are signaled in the slice header. In an alternative example, for the short-term reference pictures in the reference picture set of the coded picture, some short-term reference pictures are signaled in the referenced picture parameter set, and other short-term reference pictures are signaled in the slice header.

For reference picture list initialization, in an alternative example, short-term reference pictures may be added to the list first, and then, as appropriate, long-term reference pictures may be added to the list. After this operation, if the number of entries in the reference picture list (RefPicList0 or RefPicList1) is still less than the value of num_ref_idx_lx_active_minus1+1 (lx is 10 or 11), the remaining entries may be marked as "no reference picture." In an alternative example, the following scenario may also be possible: if the reference picture list is still incomplete after adding entries from RefPicSetStCurr0 and RefPicSetStCurr1, and possibly long-term reference pictures, pictures from RefPicSetStFoll0 and/or RefPicSetStFoll1 may be added. In an alternative example, the reference picture list initialization process may only add short-term reference pictures to the reference picture list. In this case, only long-term reference pictures may be added to the reference picture list using reference picture list modification commands, as signaled in the Reference Picture List Modification (RPLM) syntax table.

For reference picture list modifications, in an alternative example, reference picture list modifications can also be signaled differentially using ref_pic_set_idx, where the previous index is used as the predictor for the current index. In this example, different modification_of_ref_pic_idc values can correspond to different index categories (RefPicSetStCurrx for RefPicListx, RefPicSetStCurrx for RefPicList(1-x), or RefPicSetLtCurr), each of which maintains a different predictor. The predictor can be updated whenever syntax elements belonging to the same category are parsed. In an alternative example, reference picture list modifications can be based on picture number differences. In an alternative example, reference picture list modifications can be based on POC value differences.

Regarding decoded picture buffer (DPB) operations, in the example described above, after decoding the current picture and before analyzing the slice header of the next coded picture in decoding order, the current decoded picture is stored in the DPB. In an alternative example, after decoding the current picture and before analyzing the slice header of the next coded picture in decoding order, the current decoded picture is stored in temporary storage (not in the DPB). After analyzing the slice header of the next coded picture in decoding order and constructing the reference picture set for that picture, if the current decoded picture is still needed for reference or output, the current decoded picture is stored in the DPB. At this point, if the decoded picture is neither needed for reference nor output, the decoded picture can simply be discarded (from the temporary buffer).

Furthermore, in the above example, the removal of the decoded picture from the DPB occurs immediately after parsing the slice header of the current picture and before decoding any slices of the current picture. In an alternative example, the marking of the decoded picture (if any) and the removal of the decoded picture from the DPB occur after the current picture is completely decoded.

In the above example, the subsets RefPicSetStCurr0 and RefPicSetStCurr1 of the reference picture set for the current picture are derived for all decoded pictures. However, this may not be necessary for intra pictures. In an alternative example for intra picture reference picture set derivation, RefPicSetStCurr0 and RefPicSetStCurr1 are not derived for intra-coded non-IDR pictures (i.e., all slices of the coded picture are I slices) because: even if RefPicSetStCurr0 and RefPicSetStCurr1 are not empty after derivation, RefPicSetStCurr0 and RefPicSetStCurr1 are not needed in the decoding of the coded picture. Allowing non-empty RefPicSetStCurr0 or RefPicSetStCurr1 for non-IDR intra pictures may allow sharing of an example short_term_ref_pic_set() syntax structure for one or more inter-coded pictures for which both RefPicSetStCurr0 and RefPicSetStCurr1 may be non-empty.

For loss detection, the following different methods for detecting the loss of a reference picture or detecting earlier whether the current picture can be correctly decoded may be possible. In various examples, after deriving the reference picture set, video decoder 30 (e.g., the decoder side) may check for the presence of reference pictures included in RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr. If any of the reference pictures included in RefPicSetStCurr0, RefPicSetStCurr1, and RefPicSetLtCurr are not present in the DPB, the decoder side may infer that the reference picture has been lost and that the current picture will most likely not be correctly decoded, and may take some action to improve the situation, for example, by notifying the encoder side (e.g., video encoder 20) of the picture loss, and the encoder may retransmit the lost reference picture or encode the next picture(s) using only those reference pictures known to be correctly used for inter-frame prediction references at the decoder side.

In various examples, after deriving the reference picture set, the decoder side may check for the presence of reference pictures included in RefPicSetStFoll0, RefPicSetStFoll1, and RefPicSetLtFoll. If any of the reference pictures included in RefPicSetStFoll0, RefPicSetStFoll1, and RefPicSetLtFoll are not present in the DPB, the decoder side may infer that the reference picture has been lost and that some of the subsequent pictures in decoding order will likely not be decoded correctly unless some action is taken, and may take some action to remedy the situation, for example, by notifying the encoder side of the picture loss, and the encoder may retransmit the lost reference picture or encode the next picture(s) using only those reference pictures known to be correctly used for inter-prediction references at the decoder side.

For encoder-side (e.g., video encoder 20) reference picture set composition, through the above examples, the following different methods for encoder-side reference picture set composition may be possible. For example, in various examples, the encoder composes reference picture set-related syntax structures such that after deriving the reference picture set for the current picture at the decoder side: (1) RefPicSetStCurr0 includes and only includes identification information of all short-term reference pictures that have an output order earlier than the current picture and are used for reference in inter-frame prediction of the current picture, (2) RefPicSetStCurr1 includes and only includes identification information of all short-term reference pictures that have an output order later than the current picture and are used for reference in inter-frame prediction of the current picture, and (3) RefPicSetLtCurr includes and only includes identification information of all long-term reference pictures that are used for reference in inter-frame prediction of the current picture.

In various examples, an encoder (e.g., video encoder 20) may compose reference picture set-related syntax structures such that, after deriving a reference picture set for a current picture at the decoder side: (1) RefPicSetStCurr0 includes, and only includes, identification information of: 1) all short-term reference pictures that have an output order earlier than the current picture and are used for reference in inter-frame prediction of the current picture, and 2) one or more short-term reference pictures that have an output order earlier than the current picture and are not used for reference in inter-frame prediction of the current picture, (2) RefPicSetStCurr1 includes, and only includes, identification information of: 1) all short-term reference pictures that have an output order earlier than the current picture and are used for reference in inter-frame prediction of the current picture, and r1 includes and only includes identification information of the following: 1) all short-term reference pictures that have an output order later than the current picture and are used for reference in the inter-frame prediction of the current picture, and 2) one or more short-term reference pictures that have an output order later than the current picture and are not used for reference in the inter-frame prediction of the current picture, and RefPicSetLtCurr includes and only includes identification information of the following: 1) all long-term reference pictures used for reference in the inter-frame prediction of the current picture, and 2) one or more long-term reference pictures that are not used for reference in the inter-frame prediction of the current picture.

In this manner, any combination of the above-described individual techniques or individual techniques (including any combination of alternative examples) can provide techniques related to the following. However, the following list is provided for ease of understanding and should not be considered limiting. One or more of the above-described techniques can be implemented together or individually. Furthermore, the above-described techniques are examples and should not be considered limited to those specific example techniques.

Limiting the temporal_id of the reference picture set makes the DPB management method well suited for temporal scalability, may reduce signaling overhead, and may enable a simple bitstream extraction process for cleanly extracted bitstream subsets.

The long-term reference picture subset and index signaled in the picture parameter set can be included in the slice header. This can provide efficient signaling of long-term pictures.

The reference picture set is divided into various subsets, including those for the current picture, for subsequent pictures in decoding order, or for pictures with an earlier or later output order than the current picture. These scenarios can provide improved efficiency and reduced complexity for reference list initialization and reference picture list modification.

Double differential decoding in short-term image recognition signaling can provide improved efficiency. Extended and restricted long-term image recognition can provide improved efficiency and flexibility. Simplified reference picture list initialization can remove the need for "no reference picture" marking of non-completed entries in the reference picture list; however, this may not be required in all instances.

Simplified process for outputting, inserting, and removing decoded pictures from the DPB. Picture order codes (POCs) can be negative. This enables some important use cases that might not be allowed if the POC could not be negative. Signaling whether a picture is a reference picture that is not needed in the decoding process may not be necessary, but it may still be possible to signal such a situation. It may no longer be necessary to mark reference pictures as "unused for reference."

In one or more examples, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored on or transmitted via a computer-readable medium as one or more instructions or codes and executed by a hardware-based processing unit. Computer-readable media may include computer-readable storage media (which corresponds to tangible media such as data storage media) or communication media, which includes, for example, any media that facilitates the transfer of a computer program from one place to another according to a communication protocol. In this manner, computer-readable media may generally correspond to (1) non-transitory, tangible computer-readable storage media, or (2) communication media such as signals or carrier waves. Data storage media may be any available media that can be accessed by one or more computers or one or more processors to retrieve instructions, codes, and/or data structures for implementing the techniques described in this disclosure. A computer program product may include computer-readable media.

By way of example, and not limitation, such computer-readable storage media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage, flash memory, or any other medium that can be used to store desired program code in the form of instructions or data structures and that can be accessed by a computer. Furthermore, any connection may be properly termed a computer-readable medium. For example, if instructions are transmitted from a website, server, or other remote source using a coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies (e.g., infrared, radio, and microwave), then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies (e.g., infrared, radio, and microwave) are included in the definition of medium. However, it should be understood that computer-readable storage media and data storage media do not include connections, carrier waves, signals, or other transitory media, but rather refer to non-transitory, tangible storage media. As used herein, disk and disc include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk, and Blu-ray disc, where disks typically reproduce data magnetically, while discs reproduce data optically using lasers. Combinations of the above should also be included within the scope of computer-readable media.

Instructions may be executed by one or more processors, such as one or more digital signal processors (DSPs), general-purpose microprocessors, application-specific integrated circuits (ASICs), field-programmable logic arrays (FPGAs), or other equivalent integrated or discrete logic circuits. Thus, the term "processor," as used herein, may refer to any of the aforementioned structures or any other structure suitable for implementing the techniques described herein. Additionally, in some aspects, the functionality described herein may be provided within dedicated hardware and/or software modules configured for encoding and decoding, or incorporated into a combined codec. Moreover, the techniques may be fully implemented in one or more circuits or logic elements.

The techniques of this disclosure can be implemented in a wide variety of devices or apparatuses, including wireless handsets, integrated circuits (ICs), or sets of ICs (e.g., chipsets). Various components, modules, or units are described in this disclosure to emphasize functional aspects of devices configured to perform the disclosed techniques, but do not necessarily require implementation by different hardware units. Rather, as described above, the various units can be combined in a codec hardware unit, or provided by a collection of interoperable hardware units (including one or more processors as described above) in conjunction with appropriate software and/or firmware.

Various embodiments have been described. These and other embodiments are within the scope of the following claims.

Claims (20)

1. A method for encoding video data, the method comprising: The encoding indicates information about reference images belonging to a reference image set, wherein the reference image set identifies reference images that can potentially be used for inter-frame prediction of the current image and can potentially be used for inter-frame prediction of one or more images following the current image in the decoding order. Constructing a plurality of reference image subsets, each reference image subset identifying zeros or more of the reference images in the reference image set, wherein constructing the plurality of reference image subsets includes: A first reference image subset is constructed, wherein the first reference image subset identifies short-term reference images that are before the current image in the decoding order and before the current image in the output order and that can potentially be used for inter-frame prediction of one or more of the current image and the one or more images that are after the current image in the decoding order, and excludes short-term reference images that are before the current image in the output order and cannot be used for inter-frame prediction of the current image. A second subset of reference images is constructed, which identifies short-term reference images that are prior to the current image in the decoding order and follow the current image in the output order, and that can potentially be used for inter-frame prediction of one or more of the current image and the images following the current image in the decoding order; and excludes short-term reference images that are following the current image in the output order and cannot be used for inter-frame prediction of the current image; and A third reference image subset is constructed, wherein the third reference image subset identifies long-term reference images that precede the current image in the decoding order and are potentially used for inter-frame prediction of one or more of the current image and the one or more images that follow the current image in the decoding order, and excludes long-term reference images that precede the current image in the decoding order and cannot be used for inter-frame prediction of the current image. Based on which images were identified in the first, second, and third subsets of reference images, it is determined which reference images can be used for inter-frame prediction of the current image; An initial reference image list is constructed based on reference images determined to be usable for inter-frame prediction of the current image from the first, second, and third reference image subsets. Based on the need to modify the initial reference image list: Identify the reference image at the current entry in the initial reference image list; The identified reference images are added to the modified reference image list; and The current image is encoded based on the modified list of reference images. 2. The method of claim 1, wherein identifying the reference image comprises: Determine the index of at least one of the constructed reference image subsets; and The reference image identified at the entry of at least one of the constructed subset of reference images is determined based on the determined index. 3. The method of claim 2, wherein determining the index comprises: Encode a first syntax element to identify at least one of the constructed subset of reference images from which the reference image is identified; and Encode a second syntax element that indicates the index in at least one of the constructed subset of reference images. 4. The method of claim 1, further comprising: Based on the need to modify the initial reference image list: The reference image identified in the entry following the current entry in the initial reference image list is moved to the next entry to construct the modified reference image list. 5. The method of claim 1, further comprising: The encoding indicates that the reference image list modification flag needs to be modified. 6. The method of claim 1, further comprising capturing the encoded current image and the reference image. 7. An apparatus for encoding video data, the apparatus comprising: A memory configured to store a reference image; and Video encoder, the video encoder being configured to: The encoding indicates information about reference images stored in the memory that belong to a reference image set, wherein the reference image set identifies reference images that can potentially be used for inter-frame prediction of the current image and can potentially be used for inter-frame prediction of one or more images following the current image in the decoding order. Constructing multiple subsets of reference images, each subset of reference images identifies zeros or more of the reference images in the subset of reference images, wherein, to construct the multiple subsets of reference images, the video encoder is configured to: A first reference image subset is constructed, wherein the first reference image subset identifies short-term reference images that are before the current image in the decoding order and before the current image in the output order and that can potentially be used for inter-frame prediction of one or more of the current image and the one or more images that are after the current image in the decoding order, and excludes short-term reference images that are before the current image in the output order and cannot be used for inter-frame prediction of the current image. A second subset of reference images is constructed, which identifies short-term reference images that are prior to the current image in the decoding order and follow the current image in the output order, and that can potentially be used for inter-frame prediction of one or more of the current image and the images following the current image in the decoding order; and excludes short-term reference images that are following the current image in the output order and cannot be used for inter-frame prediction of the current image; and A third reference image subset is constructed, wherein the third reference image subset identifies long-term reference images that precede the current image in the decoding order and are potentially used for inter-frame prediction of one or more of the current image and the one or more images that follow the current image in the decoding order, and excludes long-term reference images that precede the current image in the decoding order and cannot be used for inter-frame prediction of the current image. Based on which images were identified in the first, second, and third subsets of reference images, it is determined which reference images can be used for inter-frame prediction of the current image; An initial reference image list is constructed based on reference images determined to be usable for inter-frame prediction of the current image from the first, second, and third reference image subsets. Based on the need to modify the initial reference image list: Identify the reference image at the current entry in the initial reference image list; The identified reference images are added to the modified reference image list; and The current image is encoded based on the modified list of reference images. 8. The apparatus of claim 7, wherein, for recognizing the reference image, the video encoder is configured to: Determine the index of at least one of the constructed reference image subsets; and The reference image identified at the entry of at least one of the constructed subset of reference images is determined based on the determined index. 9. The apparatus of claim 8, wherein, in order to determine the index, the video encoder is configured to: Encode a first syntax element to identify at least one of the constructed subset of reference images from which the reference image is identified; and Encode a second syntax element that indicates the index in at least one of the constructed subset of reference images. 10. The apparatus of claim 7, wherein the video encoder is configured to: Based on the need to modify the initial reference image list: The reference image identified in the entry following the current entry in the initial reference image list is moved to the next entry to construct the modified reference image list. 11. The apparatus of claim 7, wherein the video encoder is configured to: The encoding indicates that the reference image list modification flag needs to be modified. 12. The apparatus of claim 7, further comprising a camera configured to capture the encoded current image and the reference image. 13. The apparatus of claim 7, wherein the apparatus comprises one or more of the following: A wireless communication device comprising at least one of a display configured to display the current image or a camera configured to capture the current image; microprocessor; or integrated circuit. 14. An apparatus for encoding video data, the apparatus comprising: A means for encoding information indicating a reference image belonging to a reference image set, wherein the reference image set identifies the reference image that can potentially be used for inter-frame prediction of the current image and can potentially be used for inter-frame prediction of one or more images following the current image in the decoding order. A means for constructing a plurality of reference image subsets, each reference image subset identifying zeros or more of the reference images in the reference image set, wherein the means for constructing the plurality of reference image subsets includes: The apparatus for constructing a first subset of reference images identifies short-term reference images that precede the current image in the decoding order and precede the current image in the output order and are potentially usable for inter-frame prediction of one or more of the current image and the next one or more images that follow the current image in the decoding order, and excludes short-term reference images that precede the current image in the output order and cannot be used for inter-frame prediction of the current image. A means for constructing a second reference image subset, the second reference image subset identifying short-term reference images that are prior to the current image in the decoding order and follow the current image in the output order, and that can potentially be used for inter-frame prediction of one or more of the current image and the one or more images that are follow the current image in the decoding order, and excluding short-term reference images that are follow the current image in the output order and cannot be used for inter-frame prediction of the current image; and A means for constructing a third reference image subset, the third reference image subset identifying long-term reference images that precede the current image in the decoding order and are potentially usable for inter-frame prediction of one or more of the current image and one or more images that follow the current image in the decoding order, and excluding long-term reference images that precede the current image in the decoding order and cannot be used for inter-frame prediction of the current image. A means for determining which reference images can be used for inter-frame prediction of the current image based on which images have been identified in the first, second, and third subsets of reference images; A means for constructing an initial list of reference images based on reference images determined to be usable for inter-frame prediction of the current image from the first, second, and third subsets of reference images; A means for identifying a reference image at the current entry in the initial reference image list based on the need to modify the initial reference image list; A means for adding identified reference images to a modified reference image list based on the need to modify the initial reference image list; and A means for encoding the current image based on the modified list of reference images. 15. The apparatus of claim 14, wherein the means for identifying the reference image comprises: A means for determining an index of at least one of the constructed subsets of reference images; and A means for determining, based on a determined index, the reference image identified at an entry in at least one of the constructed subset of reference images. 16. The apparatus of claim 15, wherein the means for determining the index comprises: Means for encoding a first syntax element to identify at least one of the constructed subset of reference images from which the reference image is identified; and A means for encoding a second syntax element, the second syntax element indicating an index in at least one of the constructed subset of reference images. 17. The apparatus of claim 14, further comprising: A means for constructing the modified reference image list by moving a reference image identified in an entry following the current entry in the initial reference image list to the next entry based on a reference image list modification flag required after the initial reference image list is constructed. 18. The apparatus of claim 14, further comprising: A means for encoding a reference image list modification flag that indicates the need to modify the initial reference image list. 19. The apparatus of claim 14, further comprising means for capturing the encoded current image and the reference image. 20. The apparatus of claim 14, wherein the apparatus comprises one or more of the following: A device for displaying the current image; or A means for capturing the current image.