HK1220060B - Video coding techniques using asymmetric motion partitioning - Google Patents

Description

Video decoding technology using asymmetric motion segmentation

This application claims the benefit of U.S. Provisional Application No. 61/877,793, filed September 13, 2013, and U.S. Provisional Application No. 61/881,383, filed September 23, 2013, both of which are incorporated herein by reference in their entirety.

Technical Field

The present invention relates to video coding, ie, the encoding or decoding of video data.

Background Art

Digital video capabilities can be incorporated into a wide variety of devices, including digital televisions, digital live broadcast systems, wireless broadcast systems, personal digital assistants (PDAs), laptop or desktop computers, tablet computers, electronic book readers, digital cameras, 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 coding 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 such standards. By implementing such video coding techniques, video devices can more efficiently transmit, receive, encode, decode, and/or store digital video information.

Video coding techniques include 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 frame or portion of a video frame) may be partitioned into video blocks, which may also be referred to as treeblocks, 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 or temporal prediction results in a predictive block for the block to be coded. Residual data represents the pixel differences between the original block to be coded and the predictive block. An inter-coded block is encoded based on a motion vector pointing to a block of reference samples forming the prediction block and residual data indicating the difference between the coded block and the prediction block. An intra-coded block is 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

Generally, this disclosure relates to advanced codec-based three-dimensional (3D) video coding, including depth coding techniques in some examples. This disclosure describes techniques for view synthesis prediction coding when used with asymmetric motion partitioning, including block size determination. This disclosure also describes techniques for advanced motion prediction when used with asymmetric motion partitioning.

In one example of the present invention, a method of decoding video data includes receiving residual data corresponding to a block of video data, wherein the block of video data is encoded using asymmetric motion partitioning, is unidirectionally predicted using backward view synthesis prediction (BVSP), and has a size of 16x12, 12x16, 16x4, or 4x16; partitioning the block of video data into sub-blocks, each sub-block having a size of 8x4 or 4x8; deriving a corresponding disparity motion vector for each of the sub-blocks from a corresponding depth block in a depth picture corresponding to a reference picture; synthesizing a corresponding reference block for each of the sub-blocks using the corresponding derived disparity motion vectors; and decoding the block of video data by performing motion compensation on each of the sub-blocks using the residual data and the synthesized corresponding reference blocks.

In another example of the present invention, a method of encoding video data includes: generating a block of video data using asymmetric motion partitioning, wherein the block of video data is unidirectionally predicted using backward view synthesis prediction (BVSP) and has a size of 16x12, 12x16, 16x4, or 4x16; partitioning the block of video data into sub-blocks, each sub-block having a size of 8x4 or 4x8; deriving a corresponding disparity motion vector for each of the sub-blocks from a corresponding depth block in a depth picture corresponding to a reference picture; synthesizing a corresponding reference block for each of the sub-blocks using the corresponding derived disparity motion vectors; and encoding the block of video data by performing motion compensation on each of the sub-blocks using the synthesized corresponding reference blocks.

In another example of the present invention, an apparatus configured to decode video data includes a video memory configured to store information corresponding to a block of video data and one or more processors configured to: receive residual data corresponding to the block of video data, wherein the block of video data is encoded using asymmetric motion partitioning, is unidirectionally predicted using backward view synthesis prediction (BVSP), and has a size of 16x12, 12x16, 16x4, or 4x16; partition the block of video data into sub-blocks, each sub-block having a size of 8x4 or 4x8; derive a corresponding disparity motion vector for each of the sub-blocks from a corresponding depth block in a depth picture corresponding to a reference picture; synthesize a corresponding reference block for each of the sub-blocks using the corresponding derived disparity motion vectors; and decode the block of video data by performing motion compensation on each of the sub-blocks using the residual data and the synthesized corresponding reference blocks.

In another example of the present invention, an apparatus configured to decode video data includes: a device for receiving residual data corresponding to a block of video data, wherein the block of video data is encoded using asymmetric motion partitioning, is unidirectionally predicted using backward view synthesis prediction (BVSP), and has a size of 16x12, 12x16, 16x4, or 4x16; a device for partitioning the block of video data into sub-blocks, each sub-block having a size of 8x4 or 4x8; a device for deriving a corresponding disparity motion vector for each of the sub-blocks from a corresponding depth block in a depth picture corresponding to a reference picture; a device for synthesizing a corresponding reference block for each of the sub-blocks using the corresponding derived disparity motion vectors; and a device for decoding the block of video data by performing motion compensation on each of the sub-blocks using the residual data and the synthesized corresponding reference blocks.

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

BRIEF DESCRIPTION OF THE DRAWINGS

1 is a block diagram illustrating an example video encoding and decoding system that may utilize the inter-prediction techniques of this disclosure.

2 is a conceptual diagram illustrating an example decoding order for multi-view video.

3 is a conceptual diagram illustrating an example prediction structure for multi-view video.

FIG. 4 is a conceptual diagram illustrating texture and depth values for 3D video.

FIG5 is a conceptual diagram illustrating instance segmentation types.

6 is a conceptual diagram illustrating merge mode motion vector candidates.

FIG7 is a table indicating example specifications of merge candidate indices.

8 is a conceptual diagram illustrating neighboring blocks for an example disparity vector derivation process.

FIG. 9 is a conceptual diagram illustrating a process of deriving disparity vectors of neighboring blocks.

10 is a conceptual diagram illustrating four corner pixels of an 8x8 depth block.

11 is a conceptual diagram illustrating an example derivation of inter-view predicted motion vector candidates for merge/skip mode.

FIG. 12 is a table indicating an example specification of reference indexes in 3D-HEVC.

13 is a conceptual diagram illustrating an example derivation of motion vector inheritance candidates for depth coding.

14 illustrates the prediction structure of advanced residual prediction (ARP) in multi-view video coding.

15 is a conceptual diagram illustrating an example relationship among a current block, a reference block, and a motion compensated block.

FIG. 16 is a conceptual diagram illustrating inter-view motion prediction of a sub-prediction unit.

17 is a conceptual diagram illustrating the backward view synthesis prediction and motion compensation techniques of this disclosure when asymmetric motion partitioning is used.

18 is a conceptual diagram illustrating motion vector inheritance and motion compensation techniques for asymmetric motion partition sizes of 4x16 and 16x4.

19 is a block diagram illustrating an example of a video encoder that may implement the inter-prediction techniques of this disclosure.

20 is a block diagram illustrating an example of a video decoder that may implement the inter-prediction techniques of this disclosure.

21 is a flowchart illustrating an example encoding method of the present invention.

FIG22 is a flowchart illustrating another example encoding method of the present invention.

FIG23 is a flowchart illustrating another example encoding method of the present invention.

24 is a flow chart illustrating an example decoding method of the present invention.

25 is a flow chart illustrating an example decoding method of the present invention.

26 is a flow chart illustrating an example decoding method of the present invention.

DETAILED DESCRIPTION

In general, this disclosure describes techniques related to advanced codec-based 3D video coding, including coding of one or more views along with depth blocks using the 3D-HEVC (High Efficiency Video Coding) codec. Specifically, this disclosure describes techniques for further partitioning prediction units (PUs) partitioned using asymmetric motion partitioning techniques into smaller sub-blocks. The techniques of this disclosure include techniques for deriving and/or inheriting motion vectors and disparity motion vectors for sub-blocks of a PU partitioned using asymmetric motion partitioning.

FIG1 is a block diagram illustrating an example video encoding and decoding system 10 that may utilize the techniques of this disclosure. As shown in FIG1 , system 10 includes a source device 12 that provides encoded video data to be decoded at a later time by a destination device 14. Specifically, source device 12 may provide the video data to destination device 14 via a computer-readable medium 16. Source device 12 and destination device 14 may comprise any of a wide range of devices, including desktop computers, notebook (i.e., laptop) computers, tablet computers, set-top boxes, telephone handsets such as so-called "smart" phones, so-called "smart" tablets, televisions, cameras, display devices, digital media players, video game consoles, video streaming devices, or the like. In some cases, source device 12 and destination device 14 may be equipped for wireless communication.

Destination device 14 may receive the encoded video data to be decoded via computer-readable medium 16. Computer-readable medium 16 may comprise any type of medium or device capable of moving the encoded video data from source device 12 to destination device 14. In one example, computer-readable medium 16 may comprise a communication medium to enable source device 12 to transmit the encoded video data directly to destination device 14 in real time. The encoded video data may be modulated according to a communication standard (e.g., a wireless communication protocol) and transmitted to destination device 14. The communication medium may comprise any wireless or wired communication medium, such as a radio frequency (RF) spectrum or one or more physical transmission lines. The communication medium may form part of a packet network (e.g., a local area network, a wide area network, or a global network such as the Internet). The communication medium 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.

In some examples, the encoded data may be output from output interface 22 to a storage device. Similarly, the encoded data may be accessed from the storage device by the input interface. The storage device may include any of a variety of distributed or locally accessed data storage media, such as a hard drive, Blu-ray disc, DVD, CD-ROM, flash memory, volatile or non-volatile memory, or any other suitable digital storage medium for storing encoded video data. In another example, the storage device may correspond to a file server or another intermediate storage device that can hold the encoded video generated by source device 12. Destination device 14 may access the stored video data from the storage device via streaming or downloading. The file server may be any type of server capable of storing encoded video data and transmitting the encoded video data 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, or a local disk drive. Destination device 14 may access the encoded video data via any standard data connection, including an Internet connection. This may include a wireless channel (e.g., a Wi-Fi connection), a wired connection (e.g., DSL, cable modem, etc.), or a combination of both suitable for accessing encoded video data stored on a file server. The transmission of the encoded video data from the storage device may be a streaming transmission, a download transmission, or a combination thereof.

The techniques of this disclosure are not necessarily limited to wireless applications or settings. The techniques may be applied to video coding to support any of a variety of multimedia applications, such as over-the-air television broadcasts, cable television transmissions, satellite television transmissions, Internet streaming video transmissions (e.g., Dynamic Adaptive Streaming over HTTP (DASH)), digital video encoded onto data storage media, decoding of digital video stored on data storage media, 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 depth estimation unit 19, a video encoder 20, and an output interface 22. Destination device 14 includes an input interface 28, a video decoder 30, a depth image-based rendering (DIBR) unit 31, and a display device 32. In other examples, the source and destination devices may include other components or arrangements. For example, source device 12 may receive video data from an external video source 18, such as an external camera. Likewise, destination device 14 may interface with an external display device rather than including an integrated display device.

The illustrated system 10 of FIG. 1 is merely an example. The techniques of this disclosure may be performed by any digital video encoding and/or decoding device. Although the techniques of this disclosure are generally performed by a video encoding device, they may also be performed by a video encoder/decoder (often referred to as a "codec"). Furthermore, the techniques of this disclosure may also be performed by a video preprocessor. Source device 12 and destination device 14 are merely examples of such coding devices that generate coded video data for transmission to destination device 14 by source device 12. In some examples, devices 12, 14 may operate in a substantially symmetrical manner, such that each of devices 12, 14 includes video encoding and decoding components. Thus, system 10 may support one-way or two-way video transmission between video devices 12, 14, for example, for video streaming, video playback, video broadcasting, or video telephony.

Video source 18 of source device 12 may include a video capture device, such as a video camera, a video archive containing previously captured video, and/or a video feed interface for receiving video from a video content provider. As another alternative, video source 18 may generate computer graphics-based data as the source video, or a combination of live video, archived video, and computer-generated video. In some cases, 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, as mentioned above, the techniques described in this disclosure may be generally applicable to video decoding and may be applied to wireless and/or wired applications. In each case, the captured, pre-captured, or computer-generated video may be encoded by video encoder 20. The encoded video information may then be output by output interface 22 to computer-readable medium 16.

Video source 18 may provide one or more views of video data to video encoder 20. For example, video source 18 may correspond to an array of cameras, each of which has a unique horizontal position relative to the particular scene being captured. Alternatively, video source 18 may generate video data from different horizontal camera perspectives, for example using computer graphics. Depth estimation unit 19 may be configured to determine the values of depth pixels corresponding to pixels in the texture image. For example, depth estimation unit 19 may represent a sound navigation and ranging (SONAR) unit, a light detection and ranging (LIDAR) unit, or other unit capable of directly determining depth values substantially simultaneously as video data of a scene is recorded.

Additionally or alternatively, depth estimation unit 19 may be configured to indirectly calculate depth values by comparing two or more images captured at substantially the same time from different horizontal camera perspectives. By calculating horizontal disparity between substantially similar pixel values in the images, depth estimation unit 19 can approximate the depths of various objects in the scene. In some examples, depth estimation unit 19 may be functionally integrated with video source 18. For example, when video source 18 generates computer graphics images, depth estimation unit 19 may provide an actual depth map for the graphical objects, such as using the z-coordinates of pixels and objects used to reproduce texture images.

Computer-readable medium 16 may include transient media, such as wireless broadcasts or wired network transmissions, or storage media (that is, non-transitory storage media) such as a hard drive, a flash drive, a compact disc, a digital video disc, a Blu-ray disc, or other computer-readable media. In some examples, a network server (not shown) may receive encoded video data from source device 12 and provide the encoded video data to destination device 14, for example, via a network transmission. Similarly, a computing device at a media production facility (e.g., a disc stamping facility) may receive encoded video data from source device 12 and produce a disc containing the encoded video data. Thus, in various examples, computer-readable medium 16 may be understood to include one or more computer-readable media in various forms.

Input interface 28 of destination device 14 receives information from computer-readable medium 16. The information from computer-readable medium 16 may include syntax information defined by video encoder 20, which is also used by video decoder 30, including syntax elements that describe characteristics and/or processing of blocks and other coded units (e.g., GOPs). Display device 32 displays the decoded video data to a user and may include any of a variety of display devices, such as a cathode ray tube (CRT), a liquid crystal display (LCD), a plasma display, an organic light emitting diode (OLED) display, or another type of display device. In some examples, display device 32 may include a device capable of displaying two or more views simultaneously or substantially simultaneously, for example, to produce a 3D visual effect for a viewer.

DIBR unit 31 of destination device 14 can reproduce a synthesized view using texture and depth information for the decoded view received from video decoder 30. For example, DIBR unit 31 can determine the horizontal disparity of pixel data for the texture image based on the values of pixels in the corresponding depth map. DIBR unit 31 can then generate a synthesized image by shifting pixels in the texture image left or right by the determined horizontal disparity. In this manner, display device 32 can display one or more views that may correspond to decoded views and/or synthesized views in any combination. According to the techniques of this disclosure, video decoder 30 can provide original and updated precision values for depth ranges and camera parameters to DIBR unit 31, which can use the depth ranges and camera parameters to properly synthesize the view.

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 multiplexer-demultiplexer 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 multiplexer-demultiplexer 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 and decoder circuits, such as one or more 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, the 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, either of which may be integrated as part of a combined encoder/decoder (CODEC) in the respective device. A device including video encoder 20 and/or video decoder 30 may include an integrated circuit, a microprocessor, and/or a wireless communication device (e.g., a cellular telephone).

Video encoder 20 and video decoder 30 may operate according to a video coding standard, such as the High Efficiency Video Coding (HEVC) standard currently under development, and may conform to the HEVC Test Model (HM). Alternatively, video encoder 20 and video decoder 30 may operate according to other proprietary or industry standards, such as the ITU-T H.264 standard, alternatively referred to as MPEG-4 Part 10 Advanced Video Coding (AVC), or extensions of such standards, such as the MVC extension of ITU-T H.264/AVC. The latest joint draft of MVC is described in "Advanced Video Coding for Generic Audiovisual Services" (ITU-T Recommendation H.264) of March 2010. Specifically, video encoder 20 and video decoder 30 may operate according to 3D and/or multi-view coding standards, including a 3D extension of the HEVC standard (e.g., 3D-HEVC).

A draft of the HEVC standard, referred to as “HEVC Working Draft 10” or “WD10,” is described in document JCTVC-L1003v34, “High Efficiency Video Coding (HEVC) Textual Specification Draft 10 (for FDIS and Last Call)” by Bloss et al., Joint Collaboration Team on Video Coding (JCT-VC) of ITU-TSG16 WP3 and ISO/IEC JTC1/SC29/WG11, 12th Meeting, Geneva, Switzerland, January 14-23, 2013, which has been available for download since August 22, 2014, from http://phenix.int-evry.fr/jct/doc_end_user/documents/12_Geneva/wg11/JCTVC-L1003-v34.zip .

Another draft of the HEVC standard, referred to herein as the "WD10 revision," is described in "Editor's Proposed Corrections to HEVC Version 1" by Bloss et al., Joint Collaboration Team on Video Coding (JCT-VC) of ITU-T SG16 WP3 and ISO/IEC JTC1/SC29/WG11 , 13th Meeting, Incheon, South Korea, April 2013, available from http://phenix.int-evry.fr/jct/doc_end_user/documents/13_Incheon/wg11/JCTVC-M0432-v3.zip as of August 22, 2014. A multi-view extension to HEVC, namely MV-HEVC, is also being developed by JCT-3V.

Currently, the Joint Collaboration Team on 3D Video Coding (JCT-3C) of VCEG and MPEG is developing a 3DV standard based on HEVC. Part of this standardization effort includes the standardization of the HEVC-based Multi-View Video Codec (MV-HEVC) and another part for HEVC-based 3D Video Coding (3D-HEVC). For MV-HEVC, only the high-level syntax (HLS) changes are required, so that modules at the coding unit/prediction unit level in HEVC do not need to be redesigned and can be fully reused for MV-HEVC. For 3D-HEVC, new coding tools, including tools at the coding unit/prediction unit level, can be included and supported for both texture and depth views.

A version of the 3D-HEVC software , 3D-HTM, is available for download at the following link: [3D-HTM Version 8.0]: https://hevc.hhi.fraunhofer.de/svn/svn_3DVCSoftware/tags/HTM-8.0/ . A working draft of 3D-HEVC (Document No. E1001 ) is available at http://phenix.it-sudparis.eu/jct2/doc_end_user/current_document.php?id=1361 . The latest software description (Document No. E1005) is available at http://phenix.it-sudparis.eu/jct2/doc_end_user/current_document.php?id=1360 .

A more recent version of the 3D-HTM software for 3D-HEVC can be downloaded from the following link: [3D-HTM Version 12.0 ]: https://hevc.hhi.fraunhofer.de/svn/svn_3DVCSoftware/tags/HTM-12.0/ . The corresponding working draft of 3D-HEVC (Document No.: I1001) is available at http://phenix.int-evry.fr/jct3v/doc_end_user/current_document.php?id=2299 . The latest software description (Document No.: I1005) is available at http://phenix.int-evry.fr/jct3v/doc_end_user/current_document.php?id=2301 .

Initially, example coding techniques for HEVC will be discussed. 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 compared to existing devices, such as those based on ITU-T H.264/AVC. For example, while H.264 provides nine intra-prediction coding modes, the HM can provide up to thirty-three angular intra-prediction coding modes plus DC and planar modes.

In HEVC and other video coding specifications, a video sequence typically comprises a series of pictures. A picture may also be referred to as a "frame." A picture may comprise three sample arrays, denoted _SL , _SCb , and _SCr . _SL is a two-dimensional array (i.e., a block) of luma samples. _SCb is a two-dimensional array of Cb chroma samples. _SCr is a two-dimensional array of Cr chroma samples. Chroma samples may also be referred to herein as "chroma" samples. In other cases, a picture may be monochrome and may comprise only a luma sample array.

To generate an encoded representation of a picture, video encoder 20 may generate a set of coding tree units (CTUs). Each of the CTUs may include a coding tree block of luma samples, two corresponding coding tree blocks of chroma samples, and syntax structures for coding the samples of the coding tree blocks. In monochrome pictures or pictures with three separate color planes, a CTU may include a single coding tree block and syntax structures for coding the samples of the coding tree block. A coding tree block may be an NxN block of samples. A CTU may also be referred to as a "tree block" or a "largest coding unit (LCU)." A CTU of HEVC may be broadly similar to macroblocks of other standards, such as H.264/AVC. However, a CTU is not necessarily limited to a specific size and may include one or more coding units (CUs). A slice may include an integer number of CTUs ordered contiguously in raster scan order.

To generate a coded CTU, video encoder 20 may recursively perform quadtree partitioning on the coding tree block of the CTU to divide the coding tree block into coding blocks, hence the name "coding tree unit". A coding block is an NxN block of samples. A coding unit (CU) may include a coding block of luma samples and two corresponding coding blocks of chroma samples for a picture having a luma sample array, a Cb sample array, and a Cr sample array, and syntax structures used to code the samples of the coding blocks. In a monochrome picture or a picture with three separate color planes, a CU may include a single coding block and syntax structures used to code the samples of the coding block.

Video encoder 20 may partition the coding blocks of a CU into one or more prediction blocks. A prediction block is a rectangular (i.e., square or non-square) block of samples to which the same prediction applies. A prediction unit (PU) of a CU may include a prediction block of luma samples, two corresponding prediction blocks of chroma samples, and syntax structures used to predict the prediction blocks. In monochrome pictures or pictures with three separate color planes, a PU may include a single prediction block and syntax structures used to predict the prediction blocks. Video encoder 20 may generate predictive luma, Cb, and Cr blocks for the luma, Cb, and Cr prediction blocks of each PU of a CU.

Video encoder 20 may use intra prediction or inter prediction to generate the prediction block for the PU. If video encoder 20 uses intra prediction to generate the predictive block for the PU, video encoder 20 may generate the predictive block for the PU based on decoded samples of the picture associated with the PU. In some versions of HEVC, for the luma component of each PU, intra prediction methods are utilized with 33 angular prediction modes (indexed from 2 to 34), DC mode (indexed with 1), and planar mode (indexed with 0).

If video encoder 20 uses inter-prediction to generate the predictive block for a PU, video encoder 20 may generate the predictive block for the PU based on decoded samples of one or more pictures other than the picture associated with the PU. Inter-prediction may be unidirectional inter prediction (i.e., unidirectional prediction or unidirectional predictive prediction) or bidirectional inter prediction (i.e., bidirectional prediction or bidirectional predictive prediction). To perform unidirectional prediction or bidirectional prediction, video encoder 20 may generate a first reference picture list (RefPicList0) and a second reference picture list (RefPicList1) for the current slice. Each of the reference picture lists may include one or more reference pictures. When using unidirectional prediction, video encoder 20 may search the reference pictures in either or both of RefPicList0 and RefPicList1 to determine a reference position within the reference pictures. Furthermore, when using unidirectional prediction, video encoder 20 may generate a block of prediction samples for the PU based at least in part on samples corresponding to the reference position. Furthermore, when using unidirectional prediction, video encoder 20 may generate a single motion vector indicating the spatial displacement between the prediction block of the PU and the reference position. To indicate the spatial displacement between the PU's prediction block and the reference position, the motion vector may contain a horizontal component specifying the horizontal displacement between the PU's prediction block and the reference position and may contain a vertical component specifying the vertical displacement between the PU's prediction block and the reference position.

When encoding a PU using bi-prediction, video encoder 20 may determine a first reference location in a reference picture in RefPicList0 and a second reference location in a reference picture in RefPicList1. Video encoder 20 may then generate a predictive block for the PU based at least in part on samples corresponding to the first and second reference locations. Furthermore, when encoding a PU using bi-prediction, video encoder 20 may generate a first motion vector indicating a spatial displacement between the sample block of the PU and the first reference location, and a second motion vector indicating a spatial displacement between the prediction block of the PU and the second reference location.

Typically, reference picture list construction for the first or second reference picture list (e.g., RefPicList0 or RefPicList1) of a B picture involves two steps: reference picture list initialization and reference picture list reordering (modification). Reference picture list initialization is an explicit mechanism that places reference pictures in the reference picture memory (also known as the decoded picture buffer) into a list based on their POC (Picture Order Count) order values, which align with the display order of the pictures. The reference picture list reordering mechanism can modify the position of pictures placed in the list during reference picture list initialization to any new position, or place any reference picture in the reference picture memory at any position, even if the picture does not belong to the initialized list. Some pictures may be placed at yet another position in the list after the reference picture list reordering (modification). However, if the position of a picture exceeds the number of valid reference pictures for the list, the picture is not considered an entry in the final reference picture list. The number of valid reference pictures can be signaled in the slice header of each list.

After constructing the reference picture lists (ie, RefPicList0 and RefPicList1, if available), the reference indexes into the reference picture lists may be used to identify any reference pictures included in the reference picture lists.

After video encoder 20 generates predictive luma, Cb, and Cr blocks for one or more PUs of a CU, video encoder 20 may generate a luma residual block for the CU. Each sample in the luma residual block of the CU indicates the difference between a luma sample in one of the CU's predictive luma blocks and a corresponding sample in the CU's original luma coding block. Additionally, video encoder 20 may generate a Cb residual block for the CU. Each sample in the Cb residual block of the CU may indicate the difference between a Cb sample in one of the CU's predictive Cb blocks and a corresponding sample in the CU's original Cb coding block. Video encoder 20 may also generate a Cr residual block for the CU. Each sample in the Cr residual block of the CU may indicate the difference between a Cr sample in one of the CU's predictive Cr blocks and a corresponding sample in the CU's original Cr coding block.

In addition, video encoder 20 may use quadtree partitioning to decompose the luma, Cb, and Cr residual blocks of a CU into one or more luma, Cb, and Cr transform blocks. A transform block is a rectangular (e.g., square or non-square) block of samples to which the same transform is applied. A transform unit (TU) of a CU may include a transform block of luma samples, two corresponding transform blocks of chroma samples, and syntax structures used to transform the transform block samples. Thus, each TU of a CU may be associated with a luma transform block, a Cb transform block, and a Cr transform block. The luma transform block associated with a TU may be a subblock of the CU's luma residual block. The Cb transform block may be a subblock of the CU's Cb residual block. The Cr transform block may be a subblock of the CU's Cr residual block. In monochrome pictures or pictures with three separate color planes, a TU may include a single transform block and syntax structures used to transform the samples of the transform block.

Video encoder 20 may apply one or more transforms to the luma transform block of a TU to generate a luma coefficient block for the TU. A coefficient block may be a two-dimensional array of transform coefficients. A transform coefficient may be a scalar. Video encoder 20 may apply one or more transforms to the Cb transform block of a TU to generate a Cb coefficient block for the TU. Video encoder 20 may apply one or more transforms to the Cr transform block of a TU to generate a Cr coefficient block for the TU.

After generating a coefficient block (e.g., a luma coefficient block, a Cb coefficient block, or a Cr coefficient block), video encoder 20 may quantize the coefficient block. Quantization generally refers to the process of quantizing transform coefficients to potentially reduce the amount of data used to represent the transform coefficients, thereby providing further compression. After video encoder 20 quantizes the coefficient block, video encoder 20 may entropy encode syntax elements indicating the quantized transform coefficients. For example, video encoder 20 may perform context-adaptive binary arithmetic coding (CABAC) on the syntax elements indicating the quantized transform coefficients.

Video encoder 20 may output a bitstream comprising a sequence of bits forming a representation of a coded picture and associated data. The bitstream may comprise a series of network abstraction layer (NAL) units. A NAL unit is a syntax structure containing an indication of the type of data in the NAL unit and bytes containing the data in the form of a raw byte sequence payload (RBSP) interspersed with emulation prevention bits as needed. Each NAL unit comprises a NAL unit header and encapsulates the RBSP. The NAL unit header may include a syntax element indicating a NAL unit type code. The NAL unit type code, specified by the NAL unit header of the NAL unit, indicates the type of the NAL unit. The RBSP may be a syntax structure containing an integer number of bytes encapsulated within the NAL unit. In some cases, the RBSP comprises zero bits.

Different types of NAL units may encapsulate different types of RBSPs. For example, a first type of NAL unit may encapsulate RBSPs for picture parameter sets (PPSs), a second type of NAL unit may encapsulate RBSPs for coded slices, a third type of NAL unit may encapsulate RBSPs for SEI, and so on. NAL units that encapsulate RBSPs for video coding data (as opposed to RBSPs for parameter sets and SEI messages) may be referred to as video coding layer (VCL) NAL units.

Video decoder 30 may receive a bitstream generated by video encoder 20. Additionally, video decoder 30 may parse the bitstream to obtain syntax elements from the bitstream. Video decoder 30 may reconstruct a picture of the video data based, at least in part, on the syntax elements obtained from the bitstream. The process used to reconstruct the video data may generally be the inverse of the process performed by video encoder 20. For example, video decoder 30 may use motion vectors for PUs to determine predictive blocks for PUs of the current CU. Additionally, video decoder 30 may inverse quantize coefficient blocks associated with TUs of the current CU. Video decoder 30 may perform an inverse transform on the coefficient blocks to reconstruct transform blocks associated with the TUs of the current CU. Video decoder 30 may reconstruct the coding blocks of the current CU by adding samples of the predictive blocks for the PUs of the current CU to corresponding samples of the transform blocks for the TUs of the current CU. By reconstructing the coding blocks for each CU of a picture, video decoder 30 may reconstruct the picture.

In some examples, video encoder 20 may signal the motion information of a PU using merge mode or advanced motion vector prediction (AMVP) mode. In other words, in HEVC, there are two modes for predicting motion parameters, one is merge mode and the other is AMVP. Motion prediction may include determining the motion information of a video unit (e.g., a PU) based on the motion information of one or more other video units. The motion information of a PU may include a motion vector of the PU and a reference index of the PU.

When video encoder 20 uses merge mode to signal the motion information of the current PU, video encoder 20 generates a merge candidate list. In other words, video encoder 20 may perform a motion vector predictor list construction process. The merge candidate list includes a set of merge candidates indicating the motion information of PUs that are spatially or temporally adjacent to the current PU. That is, in merge mode, a candidate list of motion parameters (e.g., reference index, motion vector, etc.) is constructed, where the candidates may come from spatial and temporal neighboring blocks. In some examples, the candidates may also include artificially generated candidates.

Furthermore, in merge mode, video encoder 20 may select a merge candidate from a merge candidate list and may use the motion information indicated by the selected merge candidate as the motion information for the current PU. Video encoder 20 may signal the position of the selected merge candidate in the merge candidate list. For example, video encoder 20 may signal the selected motion vector parameters by transmitting an index into the candidate list. Video decoder 30 may obtain the index into the candidate list (i.e., the candidate list index) from the bitstream. Alternatively, video decoder 30 may generate the same merge candidate list and may determine the selected merge candidate based on the indication of the position of the selected merge candidate. Video decoder 30 may then use the motion information of the selected merge candidate to generate a predictive block for the current PU. That is, video decoder 30 may determine the selected candidate in the candidate list based at least in part on the candidate list index, where the selected candidate specifies the motion vector for the current PU. In this way, at the decoder side, once the index is decoded, all motion parameters for the corresponding block pointed to by the index may be inherited by the current PU.

Skip mode is similar to merge mode. In skip mode, video encoder 20 and video decoder 30 generate and use a merge candidate list in the same manner as they use a merge candidate list in merge mode. However, when video encoder 20 signals the motion information of a current PU using skip mode, video encoder 20 does not signal any residual data for the current PU. Therefore, video decoder 30 can determine the prediction block of the PU based on the reference block indicated by the motion information of the selected candidate in the merge candidate list without using the residual data.

AMVP mode is similar to merge mode in that video encoder 20 may generate a candidate list and may select a candidate from the candidate list. However, when video encoder 20 signals the RefPicListX motion information of the current PU using AMVP mode, video encoder 20 may signal the RefPicListX motion vector difference (MVD) of the current PU and the RefPicListX reference index of the current PU in addition to signaling the RefPicListX flag of the current PU. The RefPicListX MVP flag of the current PU may indicate the position of the selected AMVP candidate in the AMVP candidate list. The RefPicListX MVD of the current PU may indicate the difference between the RefPicListX motion vector of the current PU and the motion vector of the selected AMVP candidate. In this way, video encoder 20 can signal the RefPicListX motion information of the current PU by signaling the RefPicListX motion vector predictor (MVP) flag, the RefPicListX reference index value, and the RefPicListX MVD. In other words, the data representing the motion vector of the current PU in the bitstream may include data representing the reference index, an index to the candidate list, and the MVD.

In addition, when the motion information of the current PU is signaled using the AMVP mode, the video decoder 30 may obtain the MVD and MVP flag of the current PU from the bitstream. The video decoder 30 may generate the same AMVP candidate list and may determine the selected AMVP candidate based on the MVP flag. The video decoder 30 may recover the motion vector of the current PU by adding the MVD to the motion vector indicated by the selected AMVP candidate. That is, the video decoder 30 may determine the motion vector of the current PU based on the motion vector indicated by the selected AMVP candidate and the MVD. The video decoder 30 may then use the recovered one or more motion vectors of the current PU to generate a predictive block for the current PU.

When video decoder 30 generates an AMVP candidate list for a current PU, video decoder 30 may derive one or more AMVP candidates based on motion information of PUs that cover positions spatially adjacent to the current PU (i.e., spatially adjacent PUs). When the prediction block of a PU includes a position, the PU may cover the position.

A candidate in a merge candidate list or an AMVP candidate list that is based on the motion information of a PU that is temporally adjacent to the current PU (i.e., a PU at a different time instance than the current PU) may be referred to as a TMVP. That is, the TMVP may be used to improve HEVC coding efficiency, and unlike other coding tools, the TMVP may require access to the motion vectors of frames in the decoded picture buffer, more specifically, in the reference picture list.

The use of TMVP can be enabled or disabled on a per-CVS (coded video sequence) basis, per-slice basis, or on another basis. A syntax element in the SPS (e.g., sps_temporal_mvp_enable_flag) can indicate whether the use of TMVP is enabled for a CVS. Furthermore, when the use of TMVP is enabled for a CVS, the use of TMVP can be enabled or disabled for a specific slice within that CVS. For example, a syntax element in the slice header (e.g., slice_temporal_mvp_enable_flag) can indicate whether the use of TMVP is enabled for a slice. Thus, in an inter-predicted slice, when TMVP is enabled for the entire CVS (e.g., sps_temporal_mvp_enable_flag in the SPS is set to 1), a slice_temporal_mvp_enable_flag is signaled in the slice header to indicate whether TMVP is enabled for the current slice.

To determine the TMVP, the video codec may first identify a reference picture that includes a PU that is co-located with the current PU. In other words, the video coder may identify the co-located picture. If the current slice of the current picture is a B slice (i.e., a slice that allows inclusion of bi-directionally inter-predicted PUs), video encoder 20 may signal in the slice header a syntax element (e.g., collocated_from_l0_flag) that indicates whether the co-located picture is from RefPicList0 or RefPicList1. In other words, when use of TMVP is enabled for the current slice and the current slice is a B slice (e.g., a slice that allows inclusion of bi-directionally inter-predicted PUs), video encoder 20 may signal in the slice header a syntax element (e.g., collocated_from_l0_flag) that indicates whether the co-located picture is in RefPicList0 or RefPicList1. In other words, to derive the TMVP, the co-located picture will first be identified. If the current picture is a B slice, collocated_from_lO_flag is signaled in the slice header to indicate whether the co-located picture is from RefPicList0 or from RefPicList1.

After video decoder 30 identifies a reference picture list that includes a co-located picture, video decoder 30 may use another syntax element (e.g., collocated_ref_idx) that may be signaled in a slice header to identify the picture in the identified reference picture list (i.e., the co-located picture). That is, after identifying a reference picture list, collocated_ref_idx signaled in the slice header is used to identify the picture in the reference picture list.

The video coder can identify the co-located PU by examining the co-located picture. The TMVP may indicate the motion information of the bottom-right PU of the CU containing the co-located PU or the motion information of the bottom-right PU within the center PU of the CU containing the co-located PU. Therefore, the motion of the bottom-right PU of the CU containing the co-located PU or the motion of the bottom-right PU within the center PU of the CU containing the co-located PU is used. The bottom-right PU of the CU containing the co-located PU may be the PU that covers a position directly below and to the right of the bottom-right sample of the prediction block of the co-located PU. In other words, the TMVP may indicate the motion information of the PU in the reference picture that covers a position co-located with the bottom-right corner of the current PU, or the TMVP may indicate the motion information of the PU in the reference picture that covers a position co-located with the center of the current PU.

When the motion vector identified by the above process (i.e., the motion vector of the TMVP) is used to generate a motion candidate for merge mode or AMVP mode, the video coder may scale the motion vector based on the temporal position (reflected by the POC value). For example, when the difference between the POC values of the current picture and the reference picture is greater than the value when the difference between the POC values of the current picture and the reference picture is smaller, the video codec may increase the magnitude of the motion vector by the additional amount.

The target reference indexes of all possible reference picture lists for temporal merging candidates derived from TMVP may always be set to 0. However, for AMVP, the target reference indexes of all possible reference pictures are set equal to the decoded reference indexes. In other words, the target reference indexes of all possible reference picture lists for temporal merging candidates derived from TMVP are set to 0, while for AMVP, they are set equal to the decoded reference indexes. In HEVC, the SPS may include a flag (e.g., sps_temporal_mvp_enable_flag) and the slice header may include a flag (e.g., pic_temporal_mvp_enable_flag) when sps_temporal_mvp_enable_flag is equal to 1. When pic_temporal_mvp_enable_flag and temporal_id are both equal to 0 for a particular picture, motion vectors from pictures that precede the particular picture in decoding order are not used as TMVP in the decoding of the particular picture or pictures that follow the particular picture in decoding order.

In the following sections, multi-view (e.g., in H.264/MVC) and multi-view plus depth (e.g., in 3D-HEVC) coding techniques will be discussed. Initially, MVC techniques will be discussed. As described above, MVC is a multi-view coding extension of ITU-T H.264/AVC. In MVC, data for multiple views is coded in a time-first order, and therefore, the decoding order arrangement is referred to as time-first coding. Specifically, the view components (i.e., pictures) of each of multiple views at a common time instance may be coded, followed by another set of view components for a different time instance, and so on. An access unit may include coded pictures for all views for one output time instance. It should be understood that the decoding order of access units is not necessarily equivalent to the output (or display) order.

A typical MVC decoding order (i.e., bitstream order) is shown in FIG2 . This decoding order arrangement is referred to as time-first coding. Note that the decoding order of access units may not be the same as the output or display order. In FIG2 , S0 to S7 each refer to a different view of the multi-view video. T0 to T8 each represent an output time instance. An access unit may include coded pictures for all views for one output time instance. For example, a first access unit may include all views S0 to S7 for time instance T0, a second access unit may include all views S0 to S7 for time instance T1, and so on.

For the purposes of brevity, the following definitions may be used herein:

View component: A coded representation of a view in a single access unit. When a view includes both coded texture and depth representations, a view component consists of a texture view component and a depth view component.

Texture view component: A coded representation of the texture of a view in a single access unit.

Depth view component: A coded representation of the depth of a view in a single access unit.

In FIG2 , each of the views includes several picture sets. For example, view S0 includes picture sets 0, 8, 16, 24, 32, 40, 48, 56, and 64, view S1 includes picture sets 1, 9, 17, 25, 33, 41, 49, 57, and 65, and so on. For 3D video coding, such as 3D-HEVC, each picture may include two component pictures: one component picture is called a texture view component, and the other component picture is called a depth view component. Texture view components and depth view components within a picture set of a view may be considered to correspond to each other. For example, a texture view component within a set of pictures of a view is considered to correspond to a depth view component within the set of pictures of the view, and vice versa (i.e., a depth view component corresponds to its texture view component in the set, and vice versa). As used in this disclosure, a texture view component corresponding to a depth view component may be considered to be part of the same view of a single access unit.

The texture view component contains the actual image content being displayed. For example, the texture view component may include luma (Y) and chroma (Cb and Cr) components. The depth view component may indicate the relative depth of pixels in its corresponding texture view component. As an example, the depth view component is a grayscale image that only includes luma values. In other words, the depth view component may not convey any image content, but rather provides a measure of the relative depth of pixels in the texture view component.

For example, a pure white pixel in a depth view component indicates that its corresponding pixel in the corresponding texture view component is closer to the viewer's perspective, and a pure black pixel in a depth view component indicates that its corresponding pixel in the corresponding texture view component is farther from the viewer's perspective. Various shades of gray between black and white indicate different depth levels. For example, a dark gray pixel in a depth view component indicates that its corresponding pixel in the texture view component is farther away than a light gray pixel in the depth view component. Because only grayscale is needed to identify the depth of a pixel, the depth view component does not need to include chroma components, as the color values of the depth view component may not serve any purpose.

The use of only luma values (eg, intensity values) to identify depth view components is provided for purposes of illustration and should not be considered limiting. In other examples, any technique may be utilized to indicate the relative depth of pixels in a texture view component.

A typical MVC prediction structure for multi-view video coding (including both inter-picture prediction within each view and inter-view prediction) is shown in Figure 3. The prediction direction is indicated by an arrow, and the object pointed to by the arrow uses the object from which the arrow originates as a prediction reference. In MVC, inter-view prediction is supported by disparity motion compensation, which uses the syntax of H.264/AVC motion compensation but allows pictures in different views to be used as reference pictures.

3, eight views are illustrated (with view IDs "S0" through "S7"), and twelve temporal positions ("T0" through "T11") are illustrated for each view. That is, each row in FIG3 corresponds to a view, and each column indicates a temporal position.

Although MVC has so-called base views that can be decoded by H.264/AVC decoders, and MVC can also support stereoscopic view pairs, an advantage of MVC is that it can support instances where more than two views are used as 3D video input and such 3D video is decoded using multiple views. A renderer of a client with an MVC decoder can expect 3D video content with multiple views.

3 is indicated at the intersection of each row and each column. The H.264/AVC standard may use the term frame to refer to a portion of a video. This disclosure may use the terms picture and frame interchangeably.

The pictures in FIG3 are illustrated using blocks containing letters that indicate whether the corresponding picture is intra-coded (i.e., an I-picture), inter-coded in one direction (i.e., as a P-picture), or inter-coded in multiple directions (i.e., as a B-picture). Generally, prediction is indicated by an arrow, where the picture to which the arrow points uses the picture from which the arrow originates for prediction reference. For example, the P-picture of view S2 at temporal location T0 is predicted from the I-picture of view S0 at temporal location T0.

As with single-view video coding, pictures of a multi-view video coding video sequence can be predictively encoded relative to pictures at different temporal locations. For example, a b-picture of view S0 at temporal location T1 has an arrow pointing to it from an I-picture of view S0 at temporal location T0, indicating that the b-picture is predicted from the I-picture. However, in addition, with multi-view video coding, pictures can be inter-view predicted. That is, a view component can use view components in other views for reference. For example, in MVC, inter-view prediction is achieved as if a view component in another view is an inter-prediction reference. Potential inter-view references are signaled in the Sequence Parameter Set (SPS) MVC extension and can be modified through the reference picture list construction process, which enables flexible ordering of inter-prediction or inter-view prediction references. Inter-view prediction is also a feature of proposed multi-view extensions of HEVC, including 3D-HEVC (Multi-view Plus Depth).

FIG3 provides various examples of inter-view prediction. In the example of FIG3, pictures of view S1 are illustrated as being predicted from pictures at different temporal locations of view S1, as well as being inter-view predicted from pictures of views S0 and S2 at the same temporal location. For example, the b-picture of view S1 at temporal location T1 is predicted from each of the B-pictures of view S1 at temporal locations T0 and T2, as well as the b-pictures of views S0 and S2 at temporal location T1.

In some examples, FIG3 can be viewed as illustrating texture view components. For example, the I, P, B, and b pictures illustrated in FIG2 can be viewed as texture view components for each of the views. According to the techniques described in this disclosure, for each of the texture view components illustrated in FIG3, there is a corresponding depth view component. In some examples, the depth view component can be predicted in a manner similar to that illustrated in FIG3 for the corresponding texture view components.

MVC can also support the decoding of two views. One of the advantages of MVC is that an MVC encoder can treat more than two views as 3D video input and an MVC decoder can decode such multi-view representations. Therefore, any renderer with an MVC decoder can expect 3D video content with more than two views.

In MVC, inter-view prediction is allowed among pictures in the same access unit (i.e., with the same temporal instance). When coding a picture in one of the non-base views, the picture can be added to the reference picture list if it is in a different view but in the same temporal instance. An inter-view reference picture can be placed in any position in the reference picture list, just like any inter-prediction reference picture. As shown in Figure 3, a view component can use view components in other views for reference purposes. In MVC, inter-view prediction is implemented as if the view component in another view is an inter-prediction reference.

In the context of multi-view video coding, there are generally two types of motion vectors. One is called a normal motion vector. It points to a temporal reference picture and its corresponding temporal inter-frame prediction is motion compensated prediction (MCP). The other motion vector is a disparity motion vector (DMV). It points to a picture in a different view (i.e., an inter-view reference picture) and its corresponding inter-frame prediction is disparity compensated prediction (DCP).

Another type of multi-view video coding format introduces the use of depth values (e.g., in 3D-HEVC). In the Multi-view Video Plus Depth (MVD) data format, which is popular for 3D television and free-viewpoint video, texture images and depth maps can be coded independently in a multi-view texture picture. FIG4 illustrates the MVD data format with a texture image and its associated per-sample depth map. The depth range can be limited to a minimum z- _nearest and maximum z- _{farthest distance} from the camera for the corresponding 3D point.

Camera parameters and depth range values can help process decoded view components before rendering on a 3D display. Therefore, a special supplemental enhancement information (SEI) message, the multiview acquisition information SEI, is defined for the current version of H.264/MVC, which contains information specifying various parameters of the acquisition environment. However, H.264/MVC does not specify a syntax for indicating depth range-related information.

Asymmetric motion partitioning (AMP) and motion compensation block size in HEVC will now be discussed. In HEVC, an inter-coded coding block can be split into one, two, or four partitions. Various shapes of such partitions are possible. An example partitioning possibility for an inter-predicted coding block is depicted in FIG5 .

The top row of partitions in FIG5 illustrates so-called symmetric partitions. An NxN partition is simply a coding block that has not been split. An N/2xN partition is a coding block that has been split into two vertical rectangular partitions. Similarly, an NxN/2 partition is a coding block that has been split into two horizontal rectangular partitions. An N/2xN/2 partition is a coding block that has been split into four equal square partitions.

The lower four partition types in Figure 5 are called asymmetric partitions and can be used for asymmetric motion partitioning (AMP) for inter prediction. One partition of the AMP mode has a height or width of N/4 and a width or height of N, respectively, and the other partition consists of the rest of the CB by having a height or width of 3N/4 and a width or height of N. Each inter-coded partition is assigned one or two motion vectors and reference picture indices (i.e., one motion vector and reference index for unidirectional prediction and two motion vectors and reference indices for bidirectional prediction). In some examples, to minimize worst-case memory bandwidth, partitions of size 4x4 are not allowed for inter prediction, and partitions of size 4x8 and 8x4 are limited to unidirectional predictive coding based on one list of predictive data.

As will be discussed in greater detail below, this disclosure describes techniques for AMP, including backward view synthesis prediction (BVSP), when used in conjunction with 3D-HEVC coding techniques.

The following describes the merge candidate list in HEVC. For example, the merge candidate list may be constructed by the following steps. For spatial merge candidates, video encoder 20 and/or video decoder 30 may derive up to four spatial motion vector candidates from five spatial neighboring blocks, as illustrated in FIG6 .

Video encoder 20 and video decoder 30 may evaluate spatially neighboring blocks in the following order: left (A1), top (B1), top right (B0), bottom left (A0), and top left (B2), as shown in FIG6 . In some examples, a pruning process may be applied to remove motion vector candidates with identical motion information (e.g., motion vector and reference index). For example, the motion vector and reference index of B1 may be compared with the motion vector and reference index of A1, the motion vector and reference index of B0 may be compared with the motion vector and reference index of B1, the motion vector and reference index of A0 may be compared with the motion vector and reference index of A1, and the motion vector and reference index of B2 may be compared with the motion vectors and reference indexes of both B1 and A1. One of the two candidates with identical motion information may then be removed from the motion vector candidate list. If four candidates already exist after the pruning process, candidate B2 is not inserted into the merge candidate list.

If enabled and available, the co-located temporal motion vector predictor (TMVP) candidate from the reference picture is added to the motion vector candidate list after the spatial motion vector candidate.

If the motion vector candidate list is incomplete (e.g., has fewer than a predetermined number of entries), one or more artificial motion vector candidates may be generated and inserted at the end of the merge candidate list. Example types of artificial motion vector candidates include a combined bi-predictive merge candidate derived only for B slices, and a zero motion vector merge candidate (or other types of artificial motion vector candidates) to provide a predetermined number of motion vector candidates if there are not enough bi-predictive merge candidates.

When the current slice is a B slice, the derivation process of combined bi-predictive merge candidates is called. For each pair of candidates that are already in the candidate list and have the necessary motion information, a combined bi-predictive motion vector candidate (with an index denoted by combIdx) is derived using a combination of the motion vector of the first candidate (with a merge candidate index equal to 10CandIdx) of the picture in reference list 0 (if available) and the motion vector of the second candidate (with a merge candidate index equal to 11CandIdx) of the picture in reference list 1 (if available and the reference picture or motion vector is different from the first candidate).

FIG7 is a table indicating an example specification of l0CandIdx and l1CandIdx in 3D-HEVC. For example, FIG7 illustrates the definition of l0CandIdx and l1CandIdx corresponding to combIdx.

For combIdx 0…11, the process of generating combined bi-predictive motion vector candidates terminates when the following conditions are true: (1) combIdx is equal to (numOrigMergeCand*(numOrigMergeCand-1)), where numOrigMergeCand represents the number of candidates in the merge list before calling this process; (2) the total number of candidates in the merge list (including the newly generated combined bi-predictive merge candidate) is equal to MaxNumMergeCand.

This section describes the derivation of zero motion vector merge candidates. For each candidate, the zero motion vector and reference picture index are set from 0 to the number of available reference picture indices minus 1. If there are still fewer candidates than the maximum number of merge motion vector candidates (e.g., indicated by the MaxNumMergeCand syntax element), zero reference indices and motion vectors are inserted until the total number of candidates equals MaxNumMergeCand.

The following describes the constraints on motion compensation size in HEVC. To minimize the worst-case memory bandwidth, partitions of size 4x4 are not allowed for inter prediction, and partitions of size 4x8 and 8x4 are restricted to unidirectional predictive coding.

In order to satisfy this constraint mentioned above, when the current PU size is equal to 8x4 or 4x8, the generated spatial/temporal/combined bi-directional prediction merge candidates, if they are associated with bi-directional prediction mode, should reset the current PU to use uni-directional prediction by modifying the prediction direction to list 0 and the reference picture index and motion vector corresponding to RefPicList1 to -1 and (0,0) respectively.

As mentioned above, 3D-HEVC is under development. 3D-HEVC can improve coding efficiency using inter-view motion prediction and inter-view residual prediction. In other words, to further improve coding efficiency, two new technologies, inter-view motion prediction and inter-view residual prediction, have been adopted in the reference software. In inter-view motion prediction, a video coder (e.g., video encoder 20 or video decoder 30) can determine (i.e., predict) the motion information of a current PU based on the motion information of a PU in a view different from the current PU. In inter-view residual prediction, the video coder can determine a residual block for the current CU based on residual data in a view different from the current CU.

Neighboring Block-Based Disparity Vector (NBDV) derivation in 3D-HEVC will now be discussed. NBDV derivation is used as a disparity vector derivation technique in 3D-HEVC due to the fact that 3D-HEVC uses a texture-first coding order for all views. Since the corresponding depth map is not available for the currently coded texture picture, the disparity vector is derived from the neighboring blocks in NBDV. In some proposed 3D-HEVC designs, the disparity vector derived from the NBDV derivation process can be further refined by retrieving depth data corresponding to a reference texture view.

3D-HEVC initially adopted the NBDV derivation technique proposed in JCT3V-A0097 (3D-CE5.h: Disparity Vector Generation Results, L. Zhang, Y. Chen, M. Katzwirtz (Qualcomm)). JCTVC-A0126 (3D-CE5.h: Simplification of Disparity Vector Derivation for HEVC-Based 3D Video Coding (J. Sun, M. Gu, S. Ye (LG)) includes implicit disparity vectors along with simplified NBDV. In addition, in JCT3V-B0047 (3D-CE5.h related: Improvements in Disparity Vector Derivation, J. Kang, Y. Chen, L. Zhang, M. Katzwirtz (Qualcomm)), the NBDV derivation technique is further simplified by removing the implicit disparity vectors stored in the decoded picture buffer, and improving the decoding gain with random access picture (RAP) selection. Additional techniques for NBDV derivation are described in JCT3V-D0181 (CE2: CU-based Disparity Vector Derivation in 3D-HEVC, J. Kang, Y. Chen, L. Zhang, M. Katzwirtz (Qualcomm)).

The disparity vector (DV) is used as an estimator of the displacement between two views. Because neighboring blocks share almost the same motion/disparity information in video decoding, the current block can use the motion vector information in the neighboring blocks as a good predictor. Following this idea, the NBDV derivation process uses the neighboring disparity information to estimate the disparity vector in different views.

To implement NBDV, video encoder 20 initially defines several spatial and temporal neighboring blocks. Video encoder 20 then checks each of the neighboring blocks in a predefined order determined by the priority of the correlation between the current block and the candidate blocks. Once a disparity motion vector (i.e., a motion vector pointing to an inter-view reference picture) is found in a candidate, video encoder 20 converts the disparity motion vector into a disparity vector and also returns the associated view order index. Two sets of neighboring blocks are utilized. One set contains spatial neighboring blocks and the other set contains temporal neighboring blocks.

In recent proposals for 3D-HEVC, two spatially neighboring blocks are used in NBDV derivation. These are the left and top neighboring blocks relative to the current coding unit (CU) 90, as indicated by A1 and B1, respectively, in FIG8 . It should be noted that the neighboring blocks depicted in FIG8 are in the same positions as some of the neighboring blocks used in merge mode in HEVC. Therefore, no additional memory access is required. However, it should be understood that neighboring blocks in other positions relative to the current CU 90 can also be used.

To check for temporally neighboring blocks, video encoder 20 first performs a process for constructing a candidate picture list. Up to two reference pictures from the current view can be considered candidate pictures. Video encoder 20 first inserts the co-located reference picture into the candidate picture list, followed by the remaining candidate pictures in ascending order of reference index. When a reference picture with the same reference index is available in two reference picture lists, the reference picture in the same reference picture list as the co-located picture precedes the other reference picture with the same reference index. For each candidate picture in the candidate picture list, video encoder 20 determines the block covering the center position of the co-located region as a temporally neighboring block.

When a block is coded using inter-view motion prediction, a disparity vector can be derived for selecting the corresponding block in a different view. The disparity vector derived during inter-view motion prediction is called an implicit disparity vector (IDV) or also called a derived disparity vector. Even if a block is coded using motion prediction, the derived disparity vector is not discarded for the purpose of coding subsequent blocks.

In one design of HTM, during the NBDV derivation process, a video coder (e.g., video encoder 20 and/or video decoder 30) is configured to sequentially check for disparity motion vectors in temporally neighboring blocks, disparity motion vectors in spatially neighboring blocks, and then IDVs. Once a disparity motion vector or IDV is found, the process terminates.

We will now discuss the refinement of the NBDV derivation process by accessing depth information (NBDV-R). When deriving a disparity vector from the NBDV derivation process, the derived disparity vector can be further refined by retrieving depth data from the depth map of the reference view. The refinement process may include the following techniques:

a) Locate the corresponding depth block by the derived disparity vector in a previously coded reference depth view, such as the base view; the size of the corresponding depth block is the same as that of the current PU.

b) Select one depth value from the four corner pixels of the corresponding depth block and convert the depth value into the horizontal component of the refined disparity vector. The vertical component of the disparity vector remains unchanged.

Note that in some examples, the refined disparity vector can be used for inter-view motion prediction, while the unrefined disparity vector can be used for inter-view residual prediction. In addition, if the PU is coded in the backward view synthesis prediction mode, the refined disparity vector can be stored as the motion vector of the PU. In some proposals for 3D-HEVC, the depth view component of the base view can be accessed regardless of the value of the view order index derived from the NBDV derivation process.

The Backward View Synthesis Prediction (BVSP) technique in 3D-HEVC will now be discussed. An example BVSP method, "CE1.h: Backward View Synthesis Prediction Using Neighboring Blocks" (JCT3V-C0152), proposed by D. Tian et al., was adopted at the 3rd JCT-3V meeting and is available at http://phenix.it-sudparis.eu/jct2/doc_end_user/current_document.php?id=594 . The basic concept of BVSP is similar to block-based view synthesis prediction in 3D-AVC. Both techniques use backward warping and block-based view synthesis prediction to avoid transmitting motion vector differences and use more accurate motion vectors. The implementation details of BVSP in 3D-HEVC and 3D-AVC vary depending on the platform.

In 3D-HEVC, BVSP mode is supported for inter-coded blocks coded in skip or merge mode. In one example proposal for 3D-HEVC, BVSP mode is not allowed for blocks coded in advanced motion vector prediction (AMVP) mode. Instead of transmitting a flag to indicate the use of BVSP mode, video encoder 20 may be configured to add one additional merge candidate (i.e., a BVSP merge candidate) to the merge candidate list, with each candidate associated with a BVSP flag. When a decoded merge index corresponds to a BVSP merge candidate, the decoded merge index indicates that the current prediction unit (PU) uses BVSP mode. For each sub-block within the current PU, a disparity motion vector may be derived by converting the depth value in the depth reference view.

The setting of the BVSP flag is defined as follows. When the spatial neighboring block used to derive the spatial merge candidate is coded in BVSP mode, the associated motion information of the spatial merge candidate is inherited by the current block, as in normal merge mode. In addition, this spatial merge candidate is associated with a BVSP flag equal to 1 (i.e., indicating that the spatial merge candidate is coded in BVSP mode). For newly introduced BVSP merge candidates, the BVSP flag is set to 1. For all other merge candidates, the associated BVSP flag is set to 0.

As discussed above, in 3D-HEVC, video encoder 20 may be configured to derive and embed a new candidate named BVSP merge candidate into the merge candidate list.The corresponding reference picture index and motion vector are set by the following method.

The first video encoder 20 may be configured to obtain the view index represented by the view index syntax element of the derived disparity vector (e.g., refVIdxLX in 3D-HEVC) from the NBDV derivation process. Video encoder 20 may also be configured to obtain a reference picture list (e.g., RefPicListX (RefPicList0 or RefPicList1)) associated with the reference picture having a view order index equal to refVIdxLX. Video encoder 20 then uses the corresponding reference picture index and disparity vector obtained from the NBDV derivation process as motion information for the BVSP merge candidate in RefPicListX (i.e., RefPicList0 or RefPicList1).

If the current slice is a B slice, video encoder 20 checks for the availability of an inter-view reference picture with a view order index represented by refVIdxLY that is not equal to refVIdxLX in a reference picture list other than RefPicListX (i.e., RefPicListY, where Y is 1-X). If this different inter-view reference picture is found, video encoder 20 performs bi-predictive view synthesis prediction. Video encoder 20 may further be configured to use the corresponding reference picture index of the different inter-view reference picture and the scaled disparity vector from the NBDV derivation process as motion information for the BVSP merge candidate in RefPicListY. The depth block from the view with the view order index equal to refVIdxLX is used as the depth information for the current block (in the case of texture-first coding order). Video encoder 20 synthesizes the two different inter-view reference pictures (one from each reference picture list) through a backward warping process and further weights the synthesized reference pictures to arrive at the final BVSP predictor.

For slice types other than B slices (eg, P slices), video encoder 20 applies uni-predictive view synthesis prediction with RefPicListX as the reference picture list for prediction.

In 3D-HTM, texture-first coding is applied in a common test condition. Since the texture components of a view are coded before the depth components, the corresponding non-base depth component is not available when decoding a non-base texture component. Therefore, the video decoder 30 can be configured to estimate depth information and then perform BVSP using the estimated depth information. To estimate the depth information of a block, it is proposed to first derive a disparity vector from a neighboring block (e.g., using an NBDV derivation process), and then use the derived disparity vector to obtain a depth block from a reference view.

FIG9 illustrates an example technique for locating a depth block from a reference view and subsequently using the depth block for BVSP prediction. Initially, video encoder 20 and/or video decoder 30 may utilize disparity vector 104 associated with neighboring block 102. That is, video encoder 20 and/or video decoder 30 may access disparity vector information from an already encoded neighboring block (e.g., neighboring block 102) and reuse any associated disparity vector information for current block 100. Disparity vector 104 points to depth block 106 in a reference depth picture. When disparity vector 104 is reused for current block 100, disparity vector 104 now points to depth block 108 in a reference depth picture. Depth block 108 corresponds to current block 100. Video encoder 20 and/or video decoder 30 may then use the depth information in reference depth picture 108 to synthesize a block in a reference texture picture using a backward warping technique. The synthesized texture picture may then be used as a reference picture to predict current block 100.

In one example of the present disclosure, for the NBDV derivation process, it is assumed that ( _dvx , _dvy ) represents the disparity vector 104 identified by the NBDV derivation process, and the position of the current block 100 is denoted as ( _blockx , _blocky ). In one example of uni-directionally predicted BVSP, the video encoder 20 and/or video decoder 30 may be configured to retrieve a depth block 108 having a top-left position of ( _blockx + _dvx , _blocky + _dvy ) in the depth view component of the reference view. The video encoder 20 and/or video decoder 30 may be configured to first split the current block 100 (e.g., PU) into several sub-blocks, each of which has the same size (e.g., equal to W*H). For each sub-block having a size equal to W*H, the video encoder 20 and/or video decoder 30 identifies the maximum depth value of the four corner pixels of the corresponding depth sub-block 108 within the retrieved depth view component, as shown in FIG. FIG. 10 is a conceptual diagram illustrating the four corner pixels of an 8x8 depth block 110. The four corner pixels may be labeled as a top left (TL) pixel, a top right (TR) pixel, a bottom left (BL) pixel), and a bottom right (BR) pixel. Video encoder 20 and/or video decoder 30 converts the maximum depth value into a disparity motion vector. The derived disparity motion vector for each sub-block is then used for motion compensation.

This section discusses BVSP when performing bi-directional prediction. When there are multiple inter-view reference pictures from different views in RefPicList0 and RefPicList1, video encoder 20 and/or video decoder 30 applies bi-directional prediction BVSP. In bi-directional prediction BVSP, two view synthesis prediction predictors (i.e., two synthesized reference blocks) are generated from each reference list, as described above. The two view synthesis prediction predictors are then averaged to obtain a final view synthesis prediction predictor.

The motion compensation size, i.e. W*H as described above, can be 8x4 or 4x8. In one example, to determine the motion compensation size, the following rules are applied:

For each 8x8 block, check the four corners of the corresponding depth 8x8 block and:

if(vdepth[TL]<vdepth[BR]?0:1)! =(vdepth[TR]<vdepth[BL]?0:1)

Use 4x8 partitions (W=4, H=8)

else

Use 8x4 partition (W=8, H=4)

The following describes the inter-view candidate derivation process for skip/merge mode in one proposal for 3D-HEVC. Based on the disparity vector derived from the NBDV derivation process, a new motion vector candidate, called an inter-view predicted motion vector candidate (IPMVC) (if available), can be added to the AMVP and skip/merge mode motion vector candidate lists. The inter-view predicted motion vector (if available) is a temporal motion vector. Since skip mode has the same motion vector derivation process as merge mode, all techniques described in this document apply to both merge and skip modes.

FIG11 is a conceptual diagram illustrating an example derivation of inter-view predicted motion vector candidates for merge/skip mode. For example, FIG11 shows an example of a derivation process for inter-view predicted motion vector candidates. For merge/skip mode, inter-view predicted motion vector candidates are derived by the following steps. First, video encoder 20 and/or video decoder 30 uses a disparity vector to locate a corresponding block (e.g., a reference block) 112 for a current PU/CU 114 in a reference view of the same access unit. In the example of FIG11 , current block (current PU) 114 is in view V1, while corresponding reference block 112 is in view V0. If the corresponding reference block 112 is not intra-coded and not inter-view predicted, and its reference picture (in view V0 and time T1 in this example) has a POC value equal to the POC value of an entry in the same reference picture list of the current PU/CU 114, then the motion information (i.e., prediction direction, reference picture index, and motion vector) of the corresponding reference block 112 is derived as an inter-view predicted motion vector after converting the reference index based on the POC of the reference picture.

The corresponding reference block 112 can be defined as follows. First, the luma position (xP, yP) of the top left luma sample of the current prediction unit relative to the top left luma sample of the current picture is represented. The variables nPSW and nPSH represent the width and height of the current prediction unit, respectively. The reference view order index is denoted as refViewIdx, and the disparity vector is denoted as mvDisp. The reference layer luma position (xRef, yRef) is derived by the following operation:

xRef＝Clip3(0,PicWidthInSamples _L -1,xP+((nPSW-1)>>1)+((mvDisp[0]+2)>>2)) (H-124)

yRef＝Clip3(0,PicHeightInSamples _L -1,yP+((nPSH-1)>>1)+((mvDisp[1]+2)>>2)) (H-125)

The corresponding reference block 112 is set to cover the prediction unit of the luma position (xRef, yRef) in the view component with ViewIdx equal to refViewIdx.

In addition, the disparity vector can be converted into an inter-view disparity motion vector (IDMVC), which is added to the merge candidate list at a different position than the IPMVC. The inter-view disparity motion vector can also be added to the AMVP candidate list at the same position as the IPMVC (when available). IPMVC or IDMVC may be referred to as "inter-view candidate" in this context.

In one example for merge/skip mode, IPMVC (if available) is inserted into the merge candidate list before all spatial and temporal merge candidates. IDMVC is inserted before spatial merge candidates derived from _A0 .

The following section describes the construction of a merge candidate list for texture coding in 3D-HEVC. First, video encoder 20 and/or video decoder 30 derives a disparity vector, for example, using the NBDV derivation technique described above. After deriving the disparity vector, video encoder 20 and/or video decoder 30 may be configured to perform the merge candidate list construction process in 3D-HEVC as described below.

Video encoder 20 and/or video decoder 30 may derive one or more IPMVCs using the above procedure.If an IPMVC is available, the IPMV may be inserted into the merge list.

Next, video encoder 20 and/or video decoder 30 may be configured to derive spatial merging candidates and one or more IDMVC insertions in 3D-HEVC. To derive spatial merging candidates, video encoder 20 and/or video decoder 30 may be configured to check motion information of spatially neighboring PUs in the following order, for example: _A1 , _B1 , _B0 , _A0 , or _B2 , as shown in FIG.

Video encoder 20 and/or video decoder 30 may be further configured to perform constrained pruning. To perform constrained pruning, video encoder 20 and/or video decoder 30 may be configured to not insert the spatial merge candidate at position _A1 into the merge candidate list if _A1 and IPMVC have the same motion vector and the same reference index. Otherwise, the spatial merge candidate at position _A1 is inserted into the merge candidate list.

If the merge candidate at position _B1 and the merge candidate (or IPMVC) at the merge position _A1 have the same motion vector and the same reference index, the merge candidate at position _B1 is not inserted into the merge candidate list. Otherwise, the merge candidate at position _B1 is inserted into the merge candidate list. If the merge candidate at position _B0 is available (i.e., decoded and has motion information), the merge candidate at position _B0 is added to the candidate list. Video encoder 20 and/or video decoder 30 derives IDMVC using the above procedure. If IDMVC is available and its motion information is different from the candidates derived from _A1 and _B1 , the IDMVC is inserted into the candidate list.

If BVSP is enabled for the entire picture (or for the current slice), the BVSP merge candidate is inserted into the merge candidate list. If the merge candidate at position A ₀ is available, it is added to the candidate list. If the merge candidate at position B ₂ is available, it is added to the candidate list.

The next section discusses the derivation process for temporal merging candidates in 3D-HEVC. Temporal merging candidate derivation in 3D-HEVC is similar to that in HEVC, where the motion information of the co-located PU is utilized. However, for 3D-HEVC, the target reference picture index of the temporal merging candidate can be changed instead of fixing the reference picture index to 0. When the target reference index equal to 0 corresponds to a temporal reference picture (in the same view), when the motion vector of the co-located prediction unit (PU) points to an inter-view reference picture, the target reference index is changed to another index corresponding to the first entry of the inter-view reference picture in the reference picture list. Conversely, when the target reference index equal to 0 corresponds to an inter-view reference picture, when the motion vector of the co-located prediction unit (PU) points to a temporal reference picture, the target reference picture index is changed to another index corresponding to the first entry of the temporal reference picture in the reference picture list.

An example derivation process for combining bi-predictive merging candidates in 3D-HEVC will now be discussed. If the total number of candidates derived from the above two steps (i.e., derivation of spatial merging candidates and derivation of temporal merging candidates) is less than the maximum number of candidates (which may be predefined), the same process as defined in HEVC is performed, as described above. However, the specification of the reference indices l0CandIdx and l1CandIdx is different. FIG12 is another table indicating an example specification of l0CandIdx and l1CandIdx in 3D-HEVC. For example, the relationship between combIdx, l0CandIdx, and l1CandIdx is defined in the table illustrated in FIG12.

An example derivation process for zero motion vector merge candidates in 3D-HEVC is the same as that defined in HEVC. In one example of 3D-HEVC, the total number of candidates in the merge candidate list is up to 6, and a five_minus_max_num_merge_cand syntax element is generated in the slice header to specify the maximum number of merge candidates subtracted from 6. It should be noted that the value of the five_minus_max_num_merge_cand syntax element is in the range of 0 to 5, inclusive.

The following describes motion vector inheritance (MVI) for depth coding, for example, in 3D-HEVC. MVI techniques seek to exploit similarities in motion characteristics between texture components of a picture and their associated depth view components. FIG13 is a conceptual diagram illustrating an example derivation of motion vector inheritance candidates for depth coding. For example, FIG13 shows an example of an MVI candidate derivation process where the corresponding texture block 120 is selected as a 4x4 block located to the lower right of the center of the current PU 122 in the texture picture 124. For the current PU 126 in the depth picture 128, the MVI candidate reuses the motion vector and reference index associated with the already coded corresponding texture block 120 in the corresponding texture picture 124 (if such information is available).

It should be noted that motion vectors with integer precision are used in depth coding, while motion vectors with quarter precision are utilized for texture coding. Therefore, the motion vectors of the corresponding texture blocks can be scaled before being used as MVI candidates.

In case of MVI candidate generation, the merge candidate list for depth views is constructed as follows: For MVI insertion, MVI is derived using the techniques as described above and inserted into the merge candidate list if available.

The following describes the derivation process for spatial merging candidates and IDMVC insertion for depth coding in 3D-HEVC.First, video encoder 20 and/or video decoder 30 may be configured to check motion information of spatially neighboring PUs in the following order: _A1 , _B1 , _B0 , _A0 , or _B2 .

Video encoder 20 and/or video decoder 30 may then perform constrained pruning as follows. If the motion vector candidate at position _A1 and the MVI candidate have the same motion vector and the same reference index, the motion vector candidate at _A1 is not inserted into the merge candidate list. If the motion vector candidate at position _B1 and the motion vector candidate/MVI candidate at position _A1 have the same motion vector and the same reference index, the motion vector candidate at position _B1 is not inserted into the merge candidate list. If the motion vector candidate at position _B0 is available, the motion vector candidate at position _B0 is added to the merge candidate list. If the motion vector candidate at position _A0 is available, the motion vector candidate at position _A0 is added to the merge candidate list. If the motion vector candidate at position _B2 is available, the motion vector candidate at position _B2 is added to the merge candidate list.

The derivation process for temporal merging candidates in 3D-HEVC depth coding is similar to that in HEVC, where the motion information of the co-located PU is utilized. However, in 3D-HEVC depth coding, as explained above, the target reference picture index of the temporal merging candidate can be changed instead of being fixed to 0.

The derivation process for combining bi-predictive merging candidates in 3D-HEVC depth coding will now be described. If the total number of candidates derived from the above two steps is less than the maximum number of candidates, the same process as defined in HEVC is performed, except for the designation of l0CandIdx and l1CandIdx. The relationship between combIdx, l0CandIdx, and l1CandIdx is defined in the table illustrated in FIG12.

The derivation process for zero motion vector merging candidates in 3D-HEVC depth coding is the same as the procedure defined in HEVC.

The following describes an example technique for advanced residual prediction (ARP). The 4th JCT3V meeting adopted ARP applied to CUs with a partitioning mode equal to Part_2Nx2N (e.g., NxN in FIG. 5 ), as proposed in JCT3V-D0177. The JCT3V-D0177 document is Zhang et al., entitled “CE4: Advanced Residual Prediction for Multi-View Coding.” The JCT3V-D0177 document is available from http://phenix.it-sudparis.eu/jct3v/doc_end_user/current_document.php?id=862 as of August 22, 2014.

FIG14 illustrates the prediction structure of advanced residual prediction (ARP) in multi-view video coding. As shown in FIG14 , subsequent blocks are called in the prediction of a current block (“Curr”) 140. A reference block 142 in a reference/base view 144 derived from a disparity vector (DV) 146 is labeled “Base.” A block 148 in the same view (view V _m ) as the current block Curr 140, derived from a (temporal) motion vector 150 (denoted as TMV) of the current block 140, is labeled “CurrTRef.” A block 152 in the same view (view V ₀ ) as the block base 142, derived from the temporal motion vector (TMV) of the current block, is labeled “BaseTRef.” A reference block BaseTRef 152 is identified by a vector 154 of TMV+DV, which is compared to the current block Curr 140.

The residual predictor is expressed as BaseTRef-Base, where the subtraction operation is applied to each pixel of the represented pixel array. A weighting factor "w" may be further multiplied by the residual predictor. Thus, the final predictor of the current block Curr may be expressed as CurrTRef+w*(BaseTRef-Base).

Note that in the above description and FIG14 , unidirectional prediction is assumed. When ARP is extended to bidirectional prediction, the above steps are applied for each reference picture list. When the current block Curr uses an inter-view reference picture (in a different view) for one of the two reference picture lists, the ARP process is disabled.

The decoding process in ARP is described below. First, the video decoder 30 obtains the disparity vector pointing to the target reference view (for example, using the NBDV derivation process). Then, in the pictures of the reference view within the same access unit, the video decoder 30 uses the disparity vector to locate the corresponding block.

Video decoder 30 may reuse the motion information of the current block to derive motion information for the reference block.Video decoder 30 may then apply motion compensation for the corresponding block based on the same motion vector of the current block and the derived reference picture to derive a residual block.

15 is a conceptual diagram illustrating an example relationship between a current block 160, a reference block 162, and motion compensation blocks 164 and 166. A reference picture in a reference view (V ₀ ) having the same POC (Picture Order Count) value as the reference picture of the current view (V _m ) is selected as the reference picture for corresponding block 162. Video encoder 20 and/or video decoder 30 may apply a weighting factor to a residual block to obtain a weighted residual block and add the value of the weighted residual block to the prediction sample.

The weighting factors are described below. Three weighting factors are used in ARP, namely 0, 0.5, and 1. The weighting factor that results in the most minimum rate-distortion cost for the current CU is selected as the final weighting factor, and the corresponding weighting factor index (e.g., 0, 1, and 2 corresponding to weighting factors 0, 1, and 0.5, respectively) is transmitted in the bitstream at the CU level. In one example of ARP, all PU predictions in a CU share the same weighting factor. When the weighting factor is equal to 0, ARP is not used for the current CU.

The following describes some further simplifications of ARP. First, reference picture selection via motion vector scaling is described. Second, interpolation filters are described.

For reference picture selection via motion vector scaling, in JCT3V-C0049, the reference picture of a prediction unit coded with a non-zero weighting factor can differ between blocks. The JCT3V-C0049 document is Zhang et al., entitled "3D-CE4: Advanced Residual Prediction for Multiview Coding." The JCT3V-C0049 document was available on September 23, 2013, from http://phenix.int-evry.fr/jct3v/doc_end_user/current_document.php?id=487.

14 ). It has been proposed to scale the decoded motion vector of the current PU toward a fixed picture before performing motion compensation for the residual generation process when the weighting factor is not equal to 0. As proposed in JCT3V-D0177, the fixed picture is defined as the first reference picture of each reference picture list if it is from the same view. When the decoded motion vector does not point to a fixed picture, video decoder 30 may first scale the decoded motion vector and then use the scaled motion vector to identify CurrTRef and BaseTRef. This reference picture used for ARP is called a target ARP reference picture.

For interpolation filters, as described in JCT3V-C0049, video encoder 20 and/or video decoder 30 may apply a bilinear filter during the interpolation process of the corresponding block and its prediction block. For the prediction block of the current PU in a non-base view, a conventional 8/4 tap filter may be applied. In another example, as proposed in JCT3V-D0177, video encoder 20 and/or video decoder 30 may always employ bilinear filtering regardless of whether the block is in a base view or a non-base view when ARP is applied.

In one or more examples of the present disclosure, video encoder 20 and/or video decoder 30 may be configured to identify a reference view using a view order index returned from the NBDV derivation process. In one example of ARP, ARP is disabled for a reference picture list when the reference pictures of a PU in the reference picture list are from a different view than the current view.

Some additional techniques for depth inter coding are described in U.S. Provisional Application No. 61/840,400, filed on June 27, 2013, and U.S. Provisional Application No. 61/847,942, filed on July 18, 2013. In these examples, when coding a depth picture, a disparity vector is converted from neighboring samples of the current block by estimating depth values.

In other examples of ARP, additional merge candidates may be derived, eg, by accessing a reference block of the base view identified by a disparity vector.

The following describes a technique for locating blocks for inter-view motion prediction. In 3D-HEVC, two general steps are used to identify reference 4x4 blocks. The first step is to identify pixels in the reference view using disparity motion vectors. The second step is to obtain the corresponding 4x4 block (with a unique set of motion information corresponding to RefPicList0 or RefPicList1, respectively) and use that motion information to generate merge candidates.

The pixel (xRef, yRef) in the reference view is identified as follows:

xRef＝Clip3(0,PicWidthInSamples _L -1,xP+((nPSW-1)>>1)+((mvDisp[0]+2)>>2)) (H-124)

yRef＝Clip3(0,PicHeightInSamples _L -1,yP+((nPSH-1)>>1)+((mvDisp[1]+2)>>2)) (H-125)

where (xP, yP) are the coordinates of the top left sample of the current PU, mvDisp is the disparity vector and nPSWxnPSH is the size of the current PU, and _{PicWidthInSamplesL} and _{PicHeightInSamplesL} define the resolution of the picture in the reference view (same as the current view).

The following describes sub-PU-level inter-view motion prediction. JCT3V-E0184 proposes using a sub-PU-level inter-view motion prediction method for IPMVC, i.e., candidates derived from reference blocks in the reference view. An et al.'s JCT3V-E0184, "3D-CE3.h Related: Sub-PU-level Inter-view Motion Prediction," is available at http://phenix.it-sudparis.eu/jct2/doc_end_user/current_document.php?id=1198.

The basic concept of inter-view motion prediction is described above (e.g., with respect to the inter-view candidate derivation process for skip/merge mode), where only the motion information of the reference block associated with the center position is used for the current PU in the dependent view. However, the current PU may correspond to a reference area in the reference view (having the same size as the current PU identified by the disparity vector), and the reference area may have rich motion information (i.e., more than the motion vector).

Therefore, a sub-PU level inter-view motion prediction (SPIVMP) method is proposed. FIG16 is a conceptual diagram illustrating sub-prediction unit (PU) inter-view motion prediction. As shown in FIG16, current PU 170 in current view V1 can be split into multiple sub-PUs (e.g., four sub-PUs). The disparity vector of each sub-PU can be used to locate the corresponding reference block in reference view V0. Video encoder 20 and/or video decoder 30 can be configured to copy (i.e., reuse) the motion vector associated with each of the reference blocks for the corresponding sub-PU of current PU 170.

In one example, temporal inter-view merge candidates are derived as follows. First, the assigned sub-PU size is represented by NxN. The current PU is first divided into multiple sub-PUs of smaller size. The size of the current PU is represented by nPSW x nPSH and the size of the sub-PU is represented by nPSWsub x nPSHSub.

nPSWsub＝min(N,nPSW)

nPSHSub＝min(N,nPSH)

Second, for each reference picture list, the default motion vector tmvLX is set to (0, 0) and the reference index refLX is set to -1. For each sub-PU in raster scan order, the following applies. DV is added to the middle position of the current sub-PU to obtain the reference sample position (xRefSub, yRefSub) as follows:

xRefSub＝Clip3(0,PicWidthInSamplesL-1,xPSub+nPSWsub/2+((mvDisp[0]+2)>>2))

yRefSub＝Clip3(0,PicHeightInSamplesL-1,yPSub+nPSHSub/2+((mvDisp[1]+2)>>2))

The block covering (xRefSub, yRefSub) in the reference view is used as the reference block of the current sub-PU.

For the identified reference block, if it is coded using a temporal motion vector, the following applies. If both refL0 and refL1 are equal to -1, and the current sub-PU is not the first in raster scan order, the motion information of the reference block is inherited by all previous sub-PUs. The associated motion parameters may be used as candidate motion parameters for the current sub-PU. The syntax elements tmvLX and refLX are updated to the motion information of the current sub-PU. Otherwise (e.g., if the reference block is intra-coded), the motion information of the current sub-PU is set to tmvLX and refLX. Different sub-PU block sizes may be applied, such as 4x4, 8x8, and 16x16. The size of the sub-PU may be signaled in the VPS.

The following describes sub-PU-level motion vector inheritance for depth coding. Similar to the proposed sub-PU-level inter-view motion prediction from one texture view to another, U.S. Provisional Application No. 61/858,089, filed on July 24, 2013, proposes a technique for applying sub-PU-level motion prediction from one texture view to the corresponding depth view. Specifically, the current PU can be partitioned into several sub-PUs, and each sub-PU uses the motion information of the co-located texture block for motion compensation. In this case, sub-PU-level MVI is supported, and the disparity vector used for inter-view motion prediction is considered to be always zero.

The current design for BVSP in 3D-HEVC exhibits the following issues. When AMP is used and the current PU size is 4x16 or 16x4, and the PU is unidirectionally predicted, BVSP is implemented by deriving a single disparity vector for the entire PU. That is, each sub-block in the PU uses the same disparity vector for reference block synthesis and motion compensation. Therefore, BVSP can be less efficient for larger block sizes because using the same disparity vector for block synthesis and motion compensation for all sub-blocks may be less optimal for some of the sub-blocks.

As another drawback, when the current PU is bidirectionally predicted, BVSP is enabled with block sizes equal to 4x8 and 8x4. However, in HEVC, motion compensation is not allowed for blocks containing less than 64 pixels (e.g., 4x8 or 8x4 blocks) (but 16x4 and 4x16 motion compensation is allowed).

In light of these limitations, this disclosure proposes techniques related to view synthesis prediction in 3D-HEVC, focusing on the size of BVSP motion compensation. According to the techniques of this disclosure, for BVSP, each PU can be split into sub-blocks, and each sub-block can be associated with a different disparity motion vector and independently motion compensated. In this way, the accuracy of BVSP can be increased for blocks partitioned with AMP, and thus, coding efficiency can be improved. According to the techniques of this disclosure, the size of sub-blocks that can be used for BVSP can be further defined as follows.

In one example of the present disclosure, when the current PU size is 16x4 (or 4x16) and the current PU is unidirectionally predicted, video encoder 20 and/or video decoder 30 may be configured to apply BVSP and motion compensation techniques to the 8x4 (or 4x8) sub-block of the current PU. That is, the size of the BVSP sub-region is 8x4 (or 4x8). Each of the sub-blocks may be assigned a disparity motion vector converted from a depth block.

FIG17 is a conceptual diagram illustrating the BVSP and motion compensation techniques of this disclosure when using AMP. In the example of FIG17 , video encoder 20 and/or video decoder 30 asymmetrically partitions current PU 250 into 4x16 blocks. It should be noted that the 4x16 partition is merely an example, and the techniques of this disclosure described with reference to FIG17 may be applied to other asymmetrical partitions, including 16x4 partitions. Video encoder 20 and/or video decoder 30 may be configured to subdivide PU 250 into 4x8 sub-blocks 255 and 256.

17 , video encoder 20 and/or video decoder 30 are configured to unidirectionally predict PU 250 using BVSP. In this regard, video encoder 20 and/or video decoder 30 may be configured to derive a disparity vector for PU 250, for example, using NBDV derivation techniques. For example, disparity vector 261 may be derived from neighboring block 252. Video encoder 20 and/or video decoder 30 may then be configured to reuse disparity vector 261 to locate a corresponding depth block 260 in a reference depth picture. According to the techniques of this disclosure, rather than using the entirety of depth block 260 to derive a disparity motion vector for PU 255, video encoder 20 and/or video decoder 30 may be configured to derive a disparity motion vector from 4x8 sub-block 264 of depth block 260 for sub-block 255, and derive a disparity motion vector from 4x8 sub-block 262 of depth block 26 for sub-block 256.

Video encoder 20 and/or video decoder 30 may then use the corresponding derived disparity motion vector to synthesize a reference block for each of sub-blocks 255 and 256 to perform motion compensation with respect to a reference picture corresponding to an inter-view reference picture having a view order equal to refVIdxLX. By deriving individual disparity motion vectors for each of sub-blocks 255 and 256, a more accurate reference view may be synthesized and an increased coding gain may be achieved for the corresponding motion compensation process.

In another example of the present disclosure, when the current PU size is 16x12 (or 12x16) and the current PU is uni-directionally predicted, video encoder 20 and/or video decoder 30 may be configured to partition the current PU into 8x4 (or 4x8) sub-blocks (also referred to as BVSP sub-regions) and derive a disparity motion vector for each sub-block using BVSP.

In another example, the size of the BVS sub-region can be assigned to 16x12 or 12x16. In yet another example, each 16x12 (or 12x16) sub-block is further partitioned into a 16x8 (or 8x16) sub-block and two 8x4 (or 4x8) sub-blocks adjacent to a 16x4 (4x16) PU in the same CU. In another example, a 16x8 (or 8x16) sub-block can be further divided into two 8x8 sub-regions or four 4x8 (or 8x4) sub-regions based on, for example, the four corners of the corresponding depth block.

In another example of the present disclosure, when both the height and width of the current PU are greater than or equal to 8 and the PU is bi-directionally predicted, video encoder 20 and/or video decoder 30 is configured to set the size of the BVSP sub-region to 8x8 instead of 4x8 or 8x4, as in previous proposals for 3D-HEVC. In another example, instead of using bi-directionally predicted BVSP for PUs having a size equal to 12x16 or 16x12, uni-directionally predicted BVSP may be applied. In this case, the motion compensation size may be further set to 4x16 or 16x4. In another example, when the current PU size is 16x4 or 4x16 and the current PU is bi-directionally predicted, the size of the BVSP sub-region is set equal to the PU size.

Sub-PU motion prediction may exhibit the following drawbacks. In this context, sub-PU motion prediction may include the sub-PU motion prediction techniques proposed in JCT3V-E0184 as described above, as well as the extension of sub-PU motion prediction to MVI from texture view to depth view.

As a drawback, when asymmetric motion partitioning (AMP) is enabled, the current PU size is equal to, for example, 4x16 or 16x4, and the sub-PU block size signaled in the VPS is equal to 8x8, these PUs will be split into two 4x8 or 8x4 sub-blocks based on previous proposals for sub-PU motion prediction. For each sub-block, motion information is inherited from the reference block. The motion information can be identified by the disparity vector in the reference texture view used for inter-view motion prediction, or it can be reused from the co-located texture block in the corresponding texture view for motion vector inheritance. In this example, bidirectional prediction based on 4x8 or 8x4 can be invoked, which is not allowed in HEVC.

As another disadvantage, when AMP is enabled, the PU size is equal to, for example, 12x16 or 16x12, and the sub-PU block size (i.e., sub-block size) signaled in the VPS is equal to 8x8, based on previous proposals for sub-PU motion prediction, these PUs will be split into two 8x8 and two 4x8/8x4 sub-blocks. Similar to the above case, the 4x8/8x4 sub-block can use bidirectional prediction, which is not allowed in HEVC.

This disclosure proposes techniques related to inter-view motion prediction and motion vector inheritance (for depth PUs). The techniques of this disclosure can be applied in the context of a merge index indicating inter-view motion prediction or MVI. Specifically, the inter-view motion prediction and/or MVI techniques of this disclosure include techniques for further partitioning an AMP PU into sub-blocks and obtaining separate motion information for each of the sub-blocks. In this way, the accuracy of inter-view motion prediction and/or MVI can be improved for each of the sub-blocks, and thus coding efficiency can be increased.

In one example of the present disclosure, when the current PU is coded using inter-view motion prediction and/or MVI and the current PU size is equal to 4x16 or 16x4, video encoder 20 and/or video decoder 30 may be configured to split the PU into two 4x8 or 8x4 sub-blocks. For each of the sub-blocks, video encoder 20 and/or video decoder 30 may be configured to obtain motion information only from the reference blocks corresponding to a specific reference picture list (e.g., RefPicList0). The motion information corresponding to the reference blocks in RefPicList0 is inherited for the 4x8 or 8x4 sub-blocks. In this case, the sub-blocks are unidirectionally predicted from the pictures in RefPicList0.

FIG18 is a conceptual diagram illustrating motion vector inheritance and motion compensation techniques for asymmetrical partitioning into PUs of size 4x16 and 16x4. For example, for a 4x16 PU, video encoder 20 and/or video decoder 30 is configured to further partition the 4x16 PU into two 4x8 sub-blocks 300 and 302. Motion information for each of sub-blocks 300 and 302 is obtained from a reference block in a reference picture belonging to a particular reference picture list (e.g., RefPicList0). Motion compensation is then performed for each of sub-blocks 300 and 302 relative to the reference block in RefPicList0. Similarly, for a 16x4 PU, video encoder 20 and/or video decoder 30 is configured to further partition the 16x5 PU into two 8x4 sub-blocks 304 and 306. Motion information for each of sub-blocks 304 and 306 is obtained from a reference block in a reference picture belonging to a particular reference picture list (e.g., RefPicList0). Motion compensation is then performed for each of sub-blocks 304 and 306 relative to a reference block in RefPicListO.

In another example of the present disclosure, when the current PU size is one of 16x12, 12x16, 4x16, or 16x4, video encoder 20 and video decoder 30 are configured not to use bi-directional prediction for 8x4/4x8 sub-blocks when sub-PU level inter-view motion prediction and/or MVI (for depth) is applied. That is, when the current PU size is one of 16x12, 12x16, 4x16, or 16x4, video encoder 20 and video decoder 30 are configured to use only uni-directional prediction for 8x4/4x8 sub-blocks when sub-PU level inter-view motion prediction and/or MVI (for depth) is applied.

In another example of the present invention, it is proposed that when sub-PU level inter-view motion prediction or MVI is applied and the current PU size is equal to 4x16 or 16x4, the PU is not split into sub-PUs.

In another example of the present invention, when sub-PU level inter-view motion prediction or MVI is applied and the current PU size is equal to 12x16 or 16x12, the PU is split into three equal-sized sub-PU blocks with a size equal to 4x16 or 16x4. For each sub-PU block, the motion information of the corresponding reference block is inherited.

In another example of the present invention, when the current PU size is equal to 12x16 or 16x12, the PU is split into two 8x8 and one 4x16 or 16x4 sub-PU blocks, where the 8x8 sub-PU forms the left or upper half of the CU containing the PU. In another aspect of this example, the 4x16 and 16x4 sub-blocks are further split into two 4x8 or 8x4 sub-PU blocks. For each 4x8 or 8x4 sub-PU, only the motion information of the reference block corresponding to the reference picture list (RefPicList0) is obtained and reused for the 4x8 or 8x4 sub-PU. In this case, the sub-PU is unidirectionally predicted from the picture in RefPicList0.

In another example of the present disclosure, when BVSP is used for a PU having a size equal to 12x16 or 16x12, the PU is split into three equally sized sub-PU blocks having a size equal to 4x16 or 16x4. Video encoder 20 and/or video decoder 30 may then derive a unique disparity motion vector for each sub-PU from the corresponding depth block.

FIG19 is a block diagram illustrating an example of a video encoder 20 that may implement the techniques of this disclosure. Video encoder 20 may perform intra- and inter-coding (including inter-view coding) of video blocks within video slices, such as slices of both texture images and depth maps. Texture information generally includes luma (brightness or intensity) and chroma (color, such as red and blue tones) information. In general, video encoder 20 may determine a coding mode relative to a luma slice and reuse prediction information from coding the luma information to encode the chroma information (e.g., by reusing partitioning information, intra-prediction mode selection, motion vectors, or the like). 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 coding modes. Inter-mode, such as uni-directional prediction (P-mode) or bi-directional prediction (B-mode), may refer to any of several temporal-based coding modes.

As shown in FIG19 , video encoder 20 receives a current video block (i.e., a block of video data, such as a luma block, a chroma block, or a depth block) within a video frame to be encoded (i.e., a texture image or a depth map). In the example of FIG19 , video encoder 20 includes video data memory 40, mode select unit 41, decoded picture buffer (DPB) 64, summer 50, transform processing unit 52, quantization unit 54, loop filter unit 63, and entropy encoding unit 56. Mode select unit 41, in turn, includes motion compensation unit 44, motion estimation unit 42, intra-prediction processing unit 46, and partitioning unit 48. For video block reconstruction, video encoder 20 also includes inverse quantization unit 58, inverse transform processing unit 60, and summer 62. Loop filter unit 63 may include a deblocking filter and an SAO filter to filter block boundaries to remove blockiness artifacts from the reconstructed video. In addition to the deblocking filter, additional filters (in-loop or post-loop) may also be used. Such filters are not shown for brevity, but may filter the output of summer 50 (as in-loop filters), if desired.

Video data memory 40 may store video data to be encoded by components of video encoder 20. The video data stored in video data memory 40 may be obtained, for example, from video source 18. DPB 64 is a buffer that stores reference video data for use by video encoder 20 in encoding video data (e.g., in intra- or inter-coding modes, also known as intra- or inter-prediction coding modes). Video data memory 40 and DPB 64 may be formed from any of a variety of memory devices, such as dynamic random access memory (DRAM) including synchronous DRAM (SDRAM), magnetoresistive RAM (MRAM), resistive RAM (RRAM), or other types of memory devices. Video data memory 40 and DPB 64 may be provided by the same memory device or separate memory devices. In various examples, video data memory 40 may be on-chip with other components of video encoder 20, or off-chip relative to those components.

During the encoding process, video encoder 20 receives a video frame or slice to be coded. The frame or slice may be divided into multiple video blocks. Motion estimation unit 42 and motion compensation unit 44 perform inter-frame predictive coding of the received video block relative to one or more blocks in one or more reference frames (including inter-view reference frames) to provide temporal prediction. Intra-frame prediction processing unit 46 may alternatively perform intra-frame predictive coding of the received video block relative to one or more neighboring blocks in the same frame or slice as the block to be coded to provide spatial prediction. Video encoder 20 may perform multiple coding passes, e.g., to select an appropriate coding mode for each block of video data.

Furthermore, partition unit 48 may partition a block of video data into sub-blocks based on evaluation of the aforementioned partitioning schemes in the aforementioned coding passes. For example, partition unit 48 may initially partition a frame or slice into LCUs and, based on rate-distortion analysis (e.g., rate-distortion optimization), partition each of the LCUs into sub-CUs. Mode select unit 41 may further generate a quadtree data structure indicating the partitioning of the LCU into sub-CUs. A leaf node CU of the quadtree may include one or more PUs and one or more TUs.

Mode select unit 41 may select one of intra- or inter-coding modes, e.g., based on the error result, and provide the resulting intra- or inter-coded block to summer 50 to generate residual block data and to summer 62 to reconstruct the encoded block for use as a reference frame. Mode select unit 41 also provides syntax elements, such as motion vectors, intra-mode indicators, partition information, and other such syntax information, to entropy encoding unit 56.

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 of video blocks. For example, a motion vector may indicate the displacement of a PU of a video block within a current video frame or picture relative to a predictive block within a reference frame (or other coded unit) relative to the current block being coded within the current frame (or other coded unit).

A predictive block is a block that is found to closely match the 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 difference metrics. In some examples, video encoder 20 may calculate values for sub-integer pixel positions of a reference picture stored in DPB 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 a motion search relative to full-pixel positions and fractional pixel positions and output a motion vector 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 with 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 DPB 64. The reference picture lists may be constructed using the techniques of this disclosure. 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 unit 42. Again, in some examples, motion estimation unit 42 and motion compensation unit 44 may be functionally integrated. Upon receiving the motion vector for the PU of the current video block, motion compensation unit 44 may locate the predictive block to which the motion vector points in one of the reference picture lists. Summer 50 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, thereby forming pixel difference values, as discussed below. Generally, motion estimation unit 42 performs motion estimation with respect to the luma component, and motion compensation unit 44 uses motion vectors calculated based on the luma component for both chroma and luma components. In this way, motion compensation unit 44 can reuse the motion information determined for the luma component to code the chroma components, so that motion estimation unit 42 does not need to perform motion search on the chroma components. Mode select unit 41 may also generate syntax elements associated with video blocks and video slices for use by video decoder 30 when decoding video blocks of the video slices.

As an alternative to the inter-prediction performed by motion estimation unit 42 and motion compensation unit 44, as described above, intra-prediction processing unit 46 may perform intra-prediction on the current block. Specifically, intra-prediction processing unit 46 may determine an intra-prediction mode to use to encode the current block. In some examples, intra-prediction processing unit 46 may encode the current block using various intra-prediction modes, e.g., during different encoding passes, and intra-prediction processing unit 46 (or mode select unit 41 in some examples) may select an appropriate intra-prediction mode to use from a test mode.

For example, intra-prediction processing unit 46 may use rate-distortion analysis to calculate rate-distortion values for various tested intra-prediction modes and select the intra-prediction mode with the best rate-distortion characteristics from among the tested modes. Rate-distortion analysis generally determines the amount of distortion (or error) between an encoded block and the original, unencoded block that was encoded to produce the encoded block, as well as the bit rate (that is, the number of bits) used to produce the encoded block. Intra-prediction processing unit 46 may calculate ratios based on the distortion and rate 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 processing unit 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. Video encoder 20 may include, in the transmitted bitstream, configuration data that may include a plurality of intra-prediction mode index tables and a plurality of modified intra-prediction mode index tables (also called codeword mapping tables), definitions of contexts for encoding various blocks, and indications of the most probable intra-prediction mode, the intra-prediction mode index table, and the modified intra-prediction mode index table to use for each of the contexts.

Video encoder 20 forms a residual video block by subtracting the prediction data from mode select unit 41 from the original video block being decoded. Summer 50 represents the one or more components that perform this subtraction operation. Transform processing unit 52 applies a transform, such as a discrete cosine transform (DCT) or a conceptually similar transform, to the residual block, producing a video block comprising residual transform coefficient values. Transform processing unit 52 may perform other transforms conceptually similar to DCT. Wavelet transforms, integer transforms, sub-band transforms, or other types of transforms may also be used. In any case, transform processing unit 52 applies the transform to the residual block, producing a block of residual transform coefficients.

The transform may convert the residual information from the pixel value domain to a transform domain (e.g., frequency domain). Transform processing unit 52 may send the resulting transform coefficients to quantization unit 54. Quantization unit 54 may quantize 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 performs entropy coding on 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 coding technique. In the case of context-based entropy coding, the context may be based on neighboring blocks. After entropy coding by entropy encoding unit 56, the encoded bitstream may be transmitted to another device (e.g., video decoder 30), or the video may be archived for later transmission or retrieval.

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

Video encoder 20 may encode depth maps in a manner substantially similar to the coding techniques used to code luma components, even in the absence of corresponding chroma components. For example, intra-prediction processing unit 46 may intra-predict blocks of depth maps, while motion estimation unit 42 and motion compensation unit 44 may inter-predict blocks of depth maps. However, as discussed above, during inter-prediction of depth maps, motion compensation unit 44 may scale (that is, adjust) the values of the reference depth map based on the difference in depth ranges and the precision values for the depth ranges. For example, if different maximum depth values in the current depth map and the reference depth map correspond to the same real-world depth, video encoder 20 may scale the maximum depth value of the reference depth map to be equal to the maximum depth value in the current depth map for prediction purposes. Additionally or alternatively, video encoder 20 may use the updated depth range values and precision values to generate view synthesis pictures for view synthesis prediction, e.g., using techniques generally similar to inter-view prediction.

As will be discussed in more detail below with reference to Figures 21-23, video encoder 20 may be configured to employ the techniques of this disclosure described above. Specifically, video encoder 20 may be configured to partition PUs into sub-blocks when the PUs are partitioned according to an asymmetric partitioning pattern. Video encoder 20 may then be configured to inherit and/or derive motion vectors or disparity motion vectors for each of the sub-blocks.

FIG20 is a block diagram illustrating an example of a video decoder 30 that may implement the techniques of this disclosure. In the example of FIG20 , video decoder 30 includes video data memory 79, entropy decoding unit 70, motion compensation unit 72, intra-prediction processing unit 74, inverse quantization unit 76, inverse transform processing unit 78, decoded picture buffer (DPB) 82, loop filter unit 83, and summer 80. In some examples, video decoder 30 may perform a decoding pass that is generally reciprocal to the encoding pass described with respect to video encoder 20 ( FIG19 ). Motion compensation unit 72 may generate prediction data based on motion vectors received from entropy decoding unit 70, while intra-prediction processing unit 74 may generate prediction data based on intra-prediction mode indicators received from entropy decoding unit 70.

Video data memory 79 can store video data, such as an encoded video bitstream, to be decoded by components of video decoder 30. The video data stored in video data memory 79 can be obtained from a local video source, such as a camera, via wired or wireless network communication of video data, or by accessing physical data storage media. Video data memory 79 can form a coded picture buffer (CPB) that stores encoded video data from the encoded video bitstream. DPB 82 is an example of a DPB that stores reference video data for use by video decoder 30 in decoding video data (e.g., in intra- or inter-frame coding modes, also known as intra- or inter-frame prediction coding modes). Video data memory 79 and DPB 82 can be formed from any of a variety of memory devices, such as dynamic random access memory (DRAM) including synchronous DRAM (SDRAM), magnetoresistive RAM (MRAM), resistive RAM (RRAM), or other types of memory devices. Video data memory 79 and DPB 82 can be provided by the same memory device or separate memory devices. In various examples, video data memory 79 may be on-chip with the other components of video decoder 30 , or off-chip relative to those components.

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 70 of video decoder 30 entropy decodes the bitstream to generate quantized coefficients, motion vectors or intra-prediction mode indicators, and other syntax elements. Entropy decoding unit 70 forwards the motion vectors and other syntax elements to motion compensation unit 72. Video decoder 30 may receive syntax elements at the video slice level and/or the video block level.

When the video slice is coded as an intra-coded (I) slice, intra-prediction processing unit 74 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 frame or picture. When the video frame is coded as an inter-coded (i.e., B, P, or GPB) slice, motion compensation unit 72 generates a predictive block for the video block of the current video slice based on the motion vectors and other syntax elements received from entropy decoding unit 70. The predictive block may be generated from one of the reference pictures within one of the reference picture lists. Video decoder 30 may construct reference frame lists: List 0 and List 1, based on the reference pictures stored in decoded picture buffer 82 using the techniques of this disclosure. Motion compensation unit 72 determines prediction information for the video block of the current video slice by parsing the motion vectors and other syntax elements, and uses the prediction information to generate a predictive block for the current video block being decoded. For example, motion compensation unit 72 uses some of the received syntax elements to determine the prediction mode (e.g., intra-prediction or inter-prediction) used to code video blocks of a video slice, the inter-prediction slice type (e.g., B slice, P slice, or GPB slice), construction information for one or more of the reference picture lists for the slice, the motion vector for each inter-coded video block of the slice, the inter-prediction status for each inter-coded video block of the slice, and other information for decoding video blocks in the current video slice.

Motion compensation unit 72 may also perform interpolation based on interpolation filters. Motion compensation unit 72 may calculate interpolated values for sub-integer pixels of a reference block using interpolation filters such as those used by video encoder 20 during encoding of the video block. In this case, motion compensation unit 72 may determine the interpolation filters used by video encoder 20 from received syntax elements and use the interpolation filters to produce the predictive block.

Inverse quantization unit 76 inverse quantizes, or dequantizes, the quantized transform coefficients provided in the bitstream and decoded by entropy decoding unit 70. The inverse quantization process may include calculating, using video decoder 30, a quantization parameter QP _Y that determines, for each video block in a video slice, a degree of quantization and, likewise, a degree of inverse quantization that should be applied.

Inverse transform processing unit 78 applies an inverse transform (e.g., an inverse DCT, an inverse integer transform, or a conceptually similar inverse transform process) to the transform coefficients in order to produce residual blocks in the pixel domain.

After motion compensation unit 72 generates a predictive block for the current video block based on the motion vector and other syntax elements, video decoder 30 forms a decoded video block by summing the residual block from inverse transform processing unit 78 with the corresponding predictive block generated by motion compensation unit 72. Summer 90 represents the one or more components that perform this summing operation. Loop filter unit 63 may include a deblocking filter and an SAO filter to filter block boundaries to remove blockiness artifacts from the reconstructed video. In addition to the deblocking filter, additional filters (in-loop or post-loop) may also be used. Such filters are not shown for brevity, but if necessary, such filters may filter the output of summer 80 (as in-loop filters). The decoded video blocks in a given frame or picture are then stored in decoded picture buffer 82, which stores reference pictures used for subsequent motion compensation. Decoded picture buffer 82 also stores the decoded video for later presentation on a display device (e.g., display device 32 of FIG. 1 ).

As will be discussed in more detail below with reference to Figures 24-26, video decoder 30 may be configured to employ the techniques of this disclosure described above. Specifically, video decoder 30 may be configured to partition PUs into sub-blocks when the PUs are partitioned according to an asymmetric partitioning pattern. Video decoder 30 may then be configured to inherit and/or derive motion vectors or disparity motion vectors for each of the sub-blocks.

21 is a flowchart illustrating an example encoding method of this disclosure. The technique of FIG. 21 may be implemented by one or more structural units of video encoder 20, such as mode select unit 41, partition unit 48, and/or motion compensation unit 44.

In one example of the present disclosure, video encoder 20 (e.g., using mode select unit 41 and partition unit 48) may be configured to generate a block of video data using AMP, wherein the block of video data is unidirectionally predicted using BVSP and has a size of 16x12, 12x16, 16x4, or 4x16 (2100). In one example of the present disclosure, the block of video data is a prediction unit.

Video encoder 20, using partition unit 48, may be further configured to partition the block of video data into sub-blocks, each sub-block having a size of 8x4 or 4x8 (2110), and derive (e.g., using motion compensation unit 44) a respective disparity motion vector for each of the sub-blocks from a corresponding depth block in a depth picture corresponding to a reference picture (2120). Video encoder 20 (e.g., using motion compensation unit 44) may be further configured to synthesize a respective reference block for each of the sub-blocks using the respective derived disparity motion vectors (2130), and encode the block of video data (e.g., using motion compensation unit 44) by performing motion compensation on each of the sub-blocks using the synthesized respective reference blocks (2140).

In another example of the disclosure, video encoder 20 may be further configured to generate one or more syntax elements indicating that the prediction unit is encoded using AMP and indicating that the prediction unit is uni-directionally predicted using BVSP, and generate a merge candidate index pointing to a BVSP candidate.

In another example of the present disclosure, video encoder 20 (e.g., using motion compensation unit 44) may be configured to derive a respective disparity motion vector for each of the sub-blocks by deriving a disparity vector for a block of video data, locating a corresponding depth block for each of the sub-blocks using the derived disparity vector, and converting one selected depth value of the corresponding depth block for each of the sub-blocks to a respective disparity motion vector.

22 is a flowchart illustrating another example encoding method of this disclosure. The technique of FIG. 22 may be implemented by one or more structural units of video encoder 20, including mode select unit 41, partition unit 48, and/or motion compensation unit 44.

In one example of the present disclosure, video encoder 20 (e.g., mode select unit 41 and partition unit 48) may be configured to generate a second block of video data using AMP, wherein the second block of video data is encoded using at least one of inter-view motion prediction or MVI and has a size of 16x4 or 4x16 (2200). Video encoder 20 (e.g., using partition unit 48) may be further configured to partition the second block of video data into sub-blocks, each sub-block having a size of 8x4 or 4x8 (2210), and derive (e.g., using motion compensation unit 44) motion information for each of the sub-blocks from a corresponding reference block (2220). Video encoder 20 may then encode the second block of video data by performing motion compensation on each of the sub-blocks using the derived motion information and a reference picture list (2230).

In another example of this disclosure, video encoder 20 (eg, using motion compensation unit 44) may be configured to perform motion compensation by performing uni-directional motion compensation relative to pictures in the one reference picture list.

23 is a flowchart illustrating another example encoding method of this disclosure. The technique of FIG. 23 may be implemented by one or more structural units of video encoder 20, such as mode select unit 41, partition unit 48, and/or motion compensation unit 44.

In one example of the present invention, video encoder 20 may be configured to generate 20 (e.g., using mode select unit 41 and partition unit 48) a second block of video data using AMP, wherein the second block of video data is encoded using at least one of inter-view motion prediction or motion vector inheritance and has a size of 16x12 or 12x16 (2300), partition the second block of video data (e.g., using partition unit 48) into a plurality of sub-blocks (2310), and encode each of the plurality of sub-blocks with uni-directional predictive prediction (e.g., using motion compensation unit 44) (2320).

24 is a flowchart illustrating an example decoding method of this disclosure. The technique of FIG. 24 may be implemented by one or more structural units of video decoder 30, such as motion compensation unit 72.

In one example of the present disclosure, video decoder 30 may be configured to receive residual data corresponding to a block of video data, wherein the block of video data is encoded using AMP, is unidirectionally predicted using BVSP, and has a size of 16x12, 12x16, 16x4, or 4x16 (2400). In one example of the present disclosure, the block of video data is a prediction unit. Video decoder 30 may further be configured to partition the block of video data into sub-blocks, each sub-block having a size of 8x4 or 4x8 (2410), and derive a respective disparity motion vector for each of the sub-blocks from a corresponding depth block in a depth picture corresponding to a reference picture (2420).

Video decoder 30 may be further configured to synthesize a respective reference block for each of the sub-blocks using the respective derived disparity motion vectors (2430), and decode the block of video data by performing motion compensation on each of the sub-blocks using the residual data and the synthesized respective reference block (2440).

In another example of the disclosure, video decoder 30 may be further configured to receive one or more syntax elements indicating that the prediction unit is encoded using asymmetric motion partitioning and indicating that the prediction unit is uni-directionally predicted using backward view synthesis prediction, and receive a merge candidate index pointing to a BVSP candidate.

In another example of the present invention, video decoder 30 may be further configured to derive a respective disparity motion vector for each of the sub-blocks by deriving a disparity vector for the block of video data, locating a corresponding depth block for each of the sub-blocks using the derived disparity vector, and converting a selected depth value of the corresponding depth block for each of the sub-blocks to a respective disparity motion vector.

25 is a flowchart illustrating an example decoding method of this disclosure. The technique of FIG. 23 may be implemented by one or more structural units of video decoder 30, such as motion compensation unit 72.

In one example of the present disclosure, video decoder 30 may be configured to receive residual data corresponding to a second block of video data, wherein the second block of video data is encoded using at least one of inter-view motion prediction or MVI and has a size of 16x4 or 4x16 (2500), partition the second block of video data into sub-blocks, each sub-block having a size of 8x4 or 4x8 (2510), derive motion information for each of the sub-blocks from a corresponding reference block (2520), and decode the second block of video data by performing motion compensation on each of the sub-blocks using the residual data, the derived motion information, and a reference picture list.

In another example of the disclosure, video decoder 30 may be further configured to perform motion compensation by performing uni-directional motion compensation with respect to pictures in the one reference picture list.

26 is a flowchart illustrating an example decoding method of this disclosure. The technique of FIG. 23 may be implemented by one or more structural units of video decoder 30, including motion compensation unit 72.

In one example of the present disclosure, video decoder 30 may be further configured to receive residual data corresponding to a second block of video data, wherein the second block of video data is encoded using at least one of inter-view motion prediction or MVI and has a size of 16x12 or 12x16 (2600), partition the second block of video data into a plurality of sub-blocks (2610), and decode each of the plurality of sub-blocks with uni-directional predictive prediction.

As explained above, the techniques of this disclosure include video encoding and decoding techniques when AMP, BVSP, inter-view motion prediction, and/or MVI are applied to blocks of video data. Specifically, the techniques of this disclosure provide more accurate coding by guiding coding techniques for sub-blocks of PUs partitioned with AMP. For example, obtaining separate disparity motion vectors for sub-blocks of a PU partitioned with AMP when the PU is coded using BVSP can increase the accuracy of view synthesis and motion prediction, and thus increase coding efficiency. As another example, obtaining separate motion information for sub-blocks of a PU partitioned with AMP when the PU is coded using inter-view motion prediction and/or MVI can increase the accuracy of motion prediction, and thus increase coding efficiency.

It should be appreciated that, depending on the example, certain actions or events of any of the techniques described herein may be performed in a different order, may be added, combined, or omitted entirely (e.g., not all described actions or events are required to practice the techniques). Furthermore, in some examples, actions or events may be performed simultaneously rather than sequentially, for example, through multithreading, interrupt processing, or multiple processors.

In one or more instances, the described functions may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored or transmitted as one or more instructions or codes on a computer-readable medium 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 that includes, for example, any media that facilitates the transfer of a computer program from one location to another according to a communication protocol. In this manner, computer-readable media may generally correspond to (1) tangible computer-readable storage media that is non-transitory, or (2) communication media, such as a signal or carrier wave. 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 can be appropriately referred to as 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 such as infrared, radio, and microwaves, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves are included in the definition of medium. However, it should be understood that the computer-readable storage media and data storage media do not include connections, carrier waves, signals, or other transient media, but rather are directed to non-transitory, tangible storage media. As used herein, disks and optical discs include compact discs (CDs), laser discs, optical discs, digital versatile discs (DVDs), floppy disks, and Blu-ray discs, where disks typically reproduce data magnetically, while optical 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. Additionally, 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 a wireless handset, an integrated circuit (IC), or a set of ICs (e.g., a chipset). Various components, modules, or units are described herein to emphasize functional aspects of a device configured to perform the disclosed techniques, but do not necessarily require implementation by different hardware units. In fact, as described above, the various units may be combined in a codec hardware unit in conjunction with appropriate software and/or firmware, or provided by a collection of interoperating hardware units, including one or more processors as described above.

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

Claims (26)

1. A method for decoding video data, the method comprising: Receive residual data corresponding to a video data block, wherein the video data block is encoded using asymmetric motion segmentation, is unidirectionally predicted using backward view synthesis prediction (BVSP), and has a size of 16x12, 12x16, 16x4, or 4x16, and wherein the video data block is a prediction unit of the decoding unit. The video data block is divided into sub-blocks, each sub-block having a size of 8x4 or 4x8; Derive the corresponding disparity motion vector for each of the sub-blocks from the corresponding depth block in the depth image corresponding to the reference image; The corresponding reference block for each of the sub-blocks is synthesized using the corresponding derived disparity motion vectors; and The video data block is decoded by performing motion compensation on each of the sub-blocks using the residual data and the synthesized corresponding reference block. 2. The method according to claim 1, further comprising: Receive an indication that the prediction unit is encoded using asymmetric motion segmentation and an indication that the prediction unit is synthesized using a backward view to predict one or more syntax elements via unidirectional prediction; and Receive the merged candidate index pointing to the BVSP candidate. 3. The method of claim 1, wherein deriving the corresponding disparity motion vector for each of the sub-blocks comprises: Export the disparity vector of the video data block; The corresponding depth block of each of the sub-blocks is located using the derived disparity vector; and A selected depth value of the corresponding depth block for each of the sub-blocks is converted into the corresponding disparity motion vector. 4. The method according to claim 1, wherein the video data block is a first video data block, the method further comprising: Receive residual data corresponding to a second video data block, wherein the second video data block is encoded using at least one of inter-view motion prediction or motion vector inheritance and has a size of 16x4 or 4x16; The second video data block is divided into sub-blocks, each sub-block having a size of 8x4 or 4x8; Derive motion information for each of the sub-blocks from a corresponding reference block; and The second video data block is decoded by performing motion compensation on each of the sub-blocks using the residual data, the derived motion information, and a list of reference images. 5. The method of claim 4, wherein performing motion compensation includes performing unidirectional motion compensation relative to the images in the list of reference images. 6. The method according to claim 1, wherein the video data block is a first video data block, the method further comprising: Receive residual data corresponding to a second video data block, wherein the second video data block is encoded using at least one of inter-view motion prediction or motion vector inheritance and has a size of 16x12 or 12x16; The second video data block is divided into multiple sub-blocks; and Each of the plurality of sub-blocks is decoded using one-way predictive prediction. 7. A method for encoding video data, the method comprising: Video data blocks are generated using asymmetric motion segmentation, wherein the video data blocks are unidirectionally predicted using backward view synthesis prediction BVSP and have a size of 16x12, 12x16, 16x4 or 4x16, and wherein the video data blocks are prediction units of the decoding unit. The video data block is divided into sub-blocks, each sub-block having a size of 8x4 or 4x8; Derive the corresponding disparity motion vector for each of the sub-blocks from the corresponding depth block in the depth image corresponding to the reference image; The corresponding reference block for each of the sub-blocks is synthesized using the corresponding derived disparity motion vectors; and The video data block is encoded by performing motion compensation on each of the sub-blocks using the synthesized corresponding reference blocks. 8. The method of claim 7, further comprising: Generate one or more syntax elements that indicate the prediction unit is encoded using asymmetric motion segmentation and that the prediction unit is synthesized using backward view and predicted unidirectionally; and Generate a merge candidate index pointing to the BVSP candidate. 9. The method of claim 7, wherein deriving the corresponding disparity motion vector for each of the sub-blocks comprises: Export the disparity vector of the video data block; The corresponding depth block of each of the sub-blocks is located using the derived disparity vector; and A selected depth value of the corresponding depth block for each of the sub-blocks is converted into the corresponding disparity motion vector. 10. The method of claim 7, wherein the video data block is a first video data block, the method further comprising: A second video data block is generated using asymmetric motion segmentation, wherein the second video data block is encoded using at least one of inter-view motion prediction or motion vector inheritance and has a size of 16x4 or 4x16. The second video data block is divided into sub-blocks, each sub-block having a size of 8x4 or 4x8; Derive motion information for each of the sub-blocks from a corresponding reference block; and The second video data block is encoded by performing motion compensation on each of the sub-blocks using the derived motion information and a list of reference images. 11. The method of claim 10, wherein performing motion compensation includes performing unidirectional motion compensation relative to the images in the list of reference images. 12. The method of claim 7, wherein the video data block is a first video data block, the method further comprising: A second video data block is generated using asymmetric motion segmentation, wherein the second video data block is encoded using at least one of inter-view motion prediction or motion vector inheritance and has a size of 16x12 or 12x16. Divide the second video data block into multiple sub-blocks; and Each of the plurality of sub-blocks is encoded using unidirectional predictive prediction. 13. An apparatus configured to decode video data, the apparatus comprising: Video storage device, configured to store information corresponding to video data blocks; and One or more processors, configured to: Receive residual data corresponding to the video data block, wherein the video data block is encoded using asymmetric motion segmentation, is predicted unidirectionally using backward view synthesis prediction (BVSP), and has a size of 16x12, 12x16, 16x4, or 4x16, and wherein the video data block is a prediction unit of the decoding unit. The video data block is divided into sub-blocks, each sub-block having a size of 8x4 or 4x8; Derive the corresponding disparity motion vector for each of the sub-blocks from the corresponding depth block in the depth image corresponding to the reference image; The corresponding reference block for each of the sub-blocks is synthesized using the corresponding derived disparity motion vectors; and The video data block is decoded by performing motion compensation on each of the sub-blocks using the residual data and the synthesized corresponding reference block. 14. The device of claim 13, wherein the one or more processors are further configured to: Receive an indication that the prediction unit is encoded using asymmetric motion segmentation and an indication that the prediction unit is synthesized using a backward view to predict one or more syntax elements via unidirectional prediction; and Receive the merged candidate index pointing to the BVSP candidate. 15. The device of claim 13, wherein the one or more processors are further configured to: Export the disparity vector of the video data block; The corresponding depth block of each of the sub-blocks is located using the derived disparity vector; and A selected depth value of the corresponding depth block for each of the sub-blocks is converted into the corresponding disparity motion vector. 16. The apparatus of claim 13, wherein the video data block is a first video data block, and wherein the one or more processors are further configured to: Receive residual data corresponding to a second video data block, wherein the second video data block is encoded using at least one of inter-view motion prediction or motion vector inheritance and has a size of 16x4 or 4x16; The second video data block is divided into sub-blocks, each sub-block having a size of 8x4 or 4x8; Derive motion information for each of the sub-blocks from a corresponding reference block; and The second video data block is decoded by performing motion compensation on each of the sub-blocks using the residual data, the derived motion information, and a list of reference images. 17. The device of claim 16, wherein the one or more processors are further configured to perform unidirectional motion compensation relative to images in the one list of reference images. 18. The apparatus of claim 13, wherein the video data block is a first video data block, and wherein the one or more processors are further configured to: Receive residual data corresponding to a second video data block, wherein the second video data block is encoded using at least one of inter-view motion prediction or motion vector inheritance and has a size of 16x12 or 12x16; The second video data block is divided into multiple sub-blocks; and Each of the plurality of sub-blocks is decoded using one-way predictive prediction. 19. The device according to claim 13, further comprising: A display configured to show the decoded video data blocks. 20. The device of claim 13, wherein the video memory and the one or more processors include a video decoder housed within one of a mobile phone, tablet computer, laptop computer, desktop computer, set-top box, or television. 21. An apparatus configured to decode video data, the apparatus comprising: A means for receiving residual data corresponding to a video data block, wherein the video data block is encoded using asymmetric motion segmentation, is unidirectionally predicted using backward view synthesis prediction (BVSP), and has a size of 16x12, 12x16, 16x4, or 4x16, and wherein the video data block is a prediction unit of a decoding unit. A means for dividing the video data block into sub-blocks, each sub-block having a size of 8x4 or 4x8; A means for deriving the corresponding disparity motion vector of each of the sub-blocks from the corresponding depth blocks in a depth image corresponding to a reference image; A means for synthesizing a corresponding reference block for each of the sub-blocks using the corresponding derived disparity motion vectors; and A means for decoding a video data block by performing motion compensation on each of the sub-blocks using the residual data and the synthesized corresponding reference block. 22. The device according to claim 21, further comprising: Means for receiving an indication that the prediction unit is encoded using asymmetric motion segmentation and an indication that the prediction unit is synthesized using backward view to predict one or more syntax elements via unidirectional prediction; and A means for receiving a merged candidate index pointing to a BVSP candidate. 23. The apparatus of claim 21, wherein the means for deriving the corresponding disparity motion vector for each of the sub-blocks comprises: Apparatus for deriving the disparity vector of the video data block; A means for locating the corresponding depth block of each of the sub-blocks using the derived disparity vector; and A means for converting a selected depth value of the corresponding depth block of each of the sub-blocks into the corresponding parallax motion vector. 24. The apparatus of claim 21, wherein the video data block is a first video data block, the apparatus further comprising: A means for receiving residual data corresponding to a second video data block, wherein the second video data block is encoded using at least one of inter-view motion prediction or motion vector inheritance and has a size of 16x4 or 4x16; A means for dividing the second video data block into sub-blocks, each sub-block having a size of 8x4 or 4x8; A means for deriving motion information of each of the sub-blocks from a corresponding reference block; and An apparatus for decoding the second video data block by performing motion compensation on each of the sub-blocks using the residual data, the derived motion information, and a list of reference images. 25. The device of claim 24, wherein the means for performing motion compensation includes means for performing unidirectional motion compensation relative to images in the list of reference images. 26. The apparatus of claim 21, wherein the video data block is a first video data block, the apparatus further comprising: A means for receiving residual data corresponding to a second video data block, wherein the second video data block is encoded using at least one of inter-view motion prediction or motion vector inheritance and has a size of 16x12 or 12x16; A means for dividing the second video data block into multiple sub-blocks; and A means for decoding each of the plurality of sub-blocks with unidirectional predictive prediction.