JP2015171114A - Moving image encoder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To immediately restore from a buffer failure in a parallel moving image encoder using a plurality of encoders.SOLUTION: When underflow of a buffer occurs in one encoder during parallel encoding processing by a plurality of encoders, information on occurrence of a buffer failure is sent to the other encoder. In the other encoder, bit stream output of a picture where the buffer failure occurs is stopped. Picture data during encoding at present is repetitively outputted to perform restoration processing of the buffer failure.

Description

  The present invention relates to the MPEG-2 system and H.264. More particularly, the present invention relates to a video encoding apparatus using the H.264 method, and in particular, an encoding method for processing high-resolution video data in parallel using a plurality of compression encoding apparatuses and multiplexing a plurality of obtained bit streams into one bit stream. And the apparatus.

  Recent advances in digital signal processing enable high-efficiency encoding of large amounts of digital information, such as moving images, still images, and audio, to be recorded on a small recording medium or transmitted via a communication medium. It has become.

  By applying such technology, a moving image encoding apparatus capable of converting a moving image taken by a television broadcast or a video camera into a stream has been developed.

  In recent years, in video equipment for business use and large televisions, the screen size of video signals has been increased to 4k2k (4096x2160 or 3840x2160) or 8k4k (7680x4320), which is larger than conventional HDTV (1920 pixels x 1080 lines), Techniques have been proposed for improving the image quality by increasing the frame rate to 120 Hz or 240 Hz.

  In general, such a moving image encoding device for professional video equipment is often encoded using only intra prediction (intra-picture) in consideration of the convenience of the editing environment.

  However, when encoding video signals with the above large screen resolution and high frame rate, the amount of information increases compared to conventional video signals, so a high-performance video encoding device capable of high-speed processing is required. The problem of becoming.

  The following prior art has been proposed as means for realizing the encoded stream output using the conventional encoding apparatus even when a video signal having a high resolution or a high frame rate is input, in order to solve the above problems. .

  In Patent Document 1, an input image is divided into a plurality of image signals, and encoding using a plurality of encoding devices is performed for each divided image.

  At this time, a configuration is disclosed in which the bit rate is assigned so that the sum of the bit rates of the respective encoding devices becomes a predetermined value.

  The target bit rate is always set to a small bit rate so that the buffer (VBV buffer) does not underflow.

  In Patent Document 2, a high frame rate video input is distributed to a plurality of encoding devices for each frame and a single bit stream is output, and a video input distributed at a half frame rate is conventionally used. A configuration is disclosed in which parallel encoding is performed in the frame rate period.

  And when multiplexing as one bit stream, it has the means to update a header information so that the frame reference relationship of each encoding apparatus may be maintained correctly.

  As described above, according to the above-described prior art, even if the input video has a higher resolution or higher frame rate than the conventional one, the processing load associated with the encoding can be distributed by parallel processing a plurality of encoding devices. Can be realized.

JP 2001-346201 A Japanese Patent No. 04388983

  However, in Patent Document 1, the target code amount is assigned to each moving image encoding device on the average with respect to the picture data code amount of the entire screen according to the number of screen divisions.

  For this reason, when only a part of the region is a complex picture with many high-frequency components, image quality degradation may be noticeable only in that region.

Further, in the case of the configuration in which the input image according to Patent Document 2 is distributed and processed in parallel, the image quality degradation in the partial area divided as described above is avoided, but the following problem newly occurs.
The problem will be described with reference to FIGS. 1 and 2 below.

  FIG. 1 illustrates an example of operation timing when encoding is performed by two encoding apparatuses, and FIG. 2 illustrates buffer transition when a buffer underflow occurs during the operation.

  As in the operation timing of FIG. 1, an input image is input in a frame period in units of one screen, and is alternately transferred to each encoding device one screen at a time.

  Therefore, for a single encoding apparatus, a maximum frame period of twice can be used as the encoding processing time.

  On the other hand, when viewed in the encoding order of the encoding target images, two picture encodings are simultaneously performed in the same frame period period as Pic 2 and 3 and Pic 3 and 4.

  For this reason, even while picture encoding is being performed by one encoder, a buffer model simulation assuming the stream output time must be performed simultaneously by the other encoder.

  Therefore, when picture data that causes the buffer to underflow in one encoding device as shown in FIG. 2 occurs (201), the other encoded picture that is being processed in parallel is also in the state before the start of encoding. The buffer model simulation result and the actual buffer occupancy amount are different, and a problem arises that the underflow continuously occurs at the timings dt1 and dt2 in the figure (202).

  In general, when a picture that causes buffer underflow occurs in a one-pass encoding device, a method for avoiding the bitstream output of the picture being discarded and replacing all prepared macroblocks with copy pictures that refer to the forward frame It has been known.

  However, since the encoding apparatus used in the above-mentioned professional video equipment encodes only with an intra picture that is highly convenient at the time of editing, there is no buffer failure avoidance measure by copy picture using frame reference prediction. There are restrictions that cannot be implemented.

  In view of the above-described points, an object of the present invention is to make it possible to immediately recover from a buffer failure in a video encoding apparatus using a plurality of encoders.

  In order to solve the above-described problems, the moving image encoding apparatus of the present invention distributes one input video signal to a plurality of encoders, and performs intra-picture encoding processing in units of pictures in each encoder in parallel. And a buffer failure detection means for detecting a buffer failure during the encoding process by the encoder, and a buffer failure detection device for outputting a single bitstream in which a plurality of bitstreams are multiplexed. A buffer failure transmitting means for transmitting the generated information to the other encoder; a buffer failure receiving means for receiving the information on the occurrence of the buffer failure; Encoding control means for performing encoding while suppressing the generated code amount, and a picture for discarding the bitstream output of encoded picture data in which a buffer failure has occurred And over data discarding means, the next picture coded picture, characterized by comprising a picture data repeat outputting means for outputting a repeated bit stream.

  According to the present invention, in a video encoding apparatus using a plurality of encoders, it is possible to immediately recover from a buffer failure.

Operation timing diagram explaining the problem of the present invention Buffer transition diagram illustrating the problem of the present invention System configuration diagram to which the present invention is applied Block diagram in the encoding apparatus according to the first embodiment of the present invention. Operation flow in encoding apparatus according to embodiment 1 of the present invention Buffer transition diagram according to Embodiment 1 of the present invention Data structure of output bitstream according to embodiment 1 of the present invention The block diagram in the encoding apparatus which concerns on Embodiment 2 of this invention. Operation flow in encoding apparatus according to Embodiment 2 of the present invention Data structure of output bitstream according to embodiment 2 of the present invention Buffer transition diagram according to Embodiment 2 of the present invention The block diagram in the encoding apparatus which concerns on Embodiment 3 of this invention. Operation flow in encoding apparatus according to Embodiment 3 of the present invention System configuration diagram according to Embodiment 4 of the present invention Operation timing chart according to Embodiment 4 of the present invention Data structure of output bitstream according to embodiment 4 of the present invention

(Embodiment 1)
<System configuration>
FIG. 3 shows a system configuration of the video encoding apparatus according to the present embodiment.

  The encoding apparatus includes a distributor 301, a first encoder 302, a frame buffer memory 303, a second encoder 304, a buffer failure transmission / reception port 305, and a multiplexer 306.

  The distributor 301 distributes the input video data to the subsequent encoding apparatus for each period obtained by dividing the input video frame rate by the number of encoding apparatuses.

  For example, an operation is performed in which 60 Hz input video is alternately transferred to the encoding devices 301 and 304 at a cycle of 30 Hz.

  The encoder 302 stores the video data for one screen transferred from the distributor 301 in the internal frame buffer memory 303, and uses the video data stored in the memory as the current image to perform encoding in the various formats described above. Do.

  Unless otherwise specified, the encoder of the present invention is an H.264 standard. It is assumed that encoding is performed using only the intra-screen prediction mode of the H.264 (AVC) system.

  Similarly, the second encoder 304 compresses and encodes the input video data and outputs encoded picture data.

  In addition, a dedicated port 305 is provided between the encoders 302 and 304 in order to transmit / receive the presence / absence of a buffer failure described later.

  In the first embodiment, the second embodiment, and the third embodiment, the operation content is described on the premise that two coding apparatuses are processed in parallel unless otherwise specified.

  The multiplexer 306 multiplexes the encoded picture data output from each of the encoders as one bit stream so as to be in the order of input video.

  The above is the system configuration in Embodiment 1 of the present invention.

<Block configuration>
Next, FIG. 4 shows internal processing blocks common to the encoders 302 and 304 having a plurality of configurations described in the system configuration.

  The encoding apparatus includes a repeat output control unit 401, a buffer failure receiving unit 402, a picture encoding unit 403, a target code amount setting unit 404, a bit stream output unit 405, and a buffer failure transmitting unit 406.

  Hereinafter, the function of each block and the control content between the blocks will be described.

  The above blocks to which the present invention is applied may be realized as dedicated hardware such as an ASIC, or as a program operating on a processor such as a CPU (not shown) provided in the encoding apparatus. It shall be good and not particularly limited.

  The repeat output control unit 401 issues, to the bit stream output unit 401, a control signal for repeatedly outputting the picture data to be encoded in the encoding device next time based on the buffer failure presence / absence notification from the buffer failure receiving unit 402. .

  The buffer failure receiving unit 402 uses a dedicated port such as an interrupt signal between the encoding devices to detect that large picture data is generated so as to cause a buffer failure in the other encoding device.

  When a buffer failure is detected, the repeat output control unit 401 and the target code amount setting unit 404 are immediately notified of the presence or absence of the failure.

  The picture encoding unit 403 performs image compression on the input image stored in the buffer memory 302 in the in-screen encoding mode, and transfers the encoded picture data to the bit stream output unit 405.

  The picture encoding 403 switches quantization control, prediction mode selection, and entropy encoding as internal operations based on the encoding parameter setting from the target code amount setting unit 404.

  When the picture data occupancy and the target generated code quantity in the buffer set in advance at the start of encoding are monitored and picture data having a data size that causes buffer underflow is encoded, this is indicated in the buffer. The failure transmitting unit 406 is notified.

  The target code amount setting unit 404 receives the buffer failure information in the other encoding device from the buffer failure receiving unit 402, the buffer failure occurrence information in the encoding from the buffer failure transmitting unit 406, and the picture encoding unit 403. Receives a generated code amount of a picture that has been encoded in units of pictures.

  Based on the above buffer failure information and the amount of code generated in the encoded picture, the buffer model occupancy position is simulated, and the target is to prevent the next encoded picture or the picture currently being encoded from overflowing or overflowing the buffer. The generated code amount is calculated and set.

  The bitstream output unit 405 receives the encoded picture data output from the picture encoding unit 403, repeats the output control signal of the picture data from the repeat output control unit 401, and the picture from the buffer failure transmission unit 406 Based on the data output stop control signal, a bit stream is output to the subsequent multiplexer 106.

  The buffer failure transmitting unit 406 receives the buffer failure generated in the picture encoding unit 403, issues a control signal for skipping the bit stream output of the picture data to the bit stream output unit 405, and the other encoding device and target The code amount setting unit 404 is notified of the occurrence of a buffer failure in the encoding apparatus.

  As the buffer failure transmitting means to the other encoding device, a means capable of instantaneously transmitting and receiving via a dedicated port by hardware, etc., as in the above-described buffer failure receiving unit 402 is desirable.

  The above is the description of the blocks constituting the encoding device to which the present invention is applied.

<Failure transmission flow>
Next, failure detection realized by using the above-described functional blocks, which is a feature of the present invention, and transmission processing to another encoding apparatus will be described with reference to each flowchart of FIG.

  In FIG. 5A, first in step S501, it is determined whether or not a buffer failure has occurred in the picture encoder 403 of the moving picture encoding apparatus.

  If no buffer failure has occurred (FALSE in step S501), the operation flow is terminated, and after the encoding of the picture is completed, the picture data is output from the bitstream output unit 405 as it is.

  On the other hand, when a buffer failure occurs (TRUE in step S501), the buffer failure transmission unit 406 immediately performs buffer failure transmission to the other encoding device (step S502).

  After the buffer failure transmission, it is determined whether or not the failure picture has been encoded, and waits until the encoding ends (step S503).

  Although not particularly limited in the present invention, it is desirable to interrupt the encoding when the occurrence of a failure is found, for example, at the timing of a macroblock unit or the like, so that the feedback control of the subsequent failure recovery process can be executed immediately.

  After encoding the failed picture (TRUE in step S503), a control signal for skipping the bit stream output of the picture data is set from the buffer failed transmitting unit 406 to the bit stream output unit 405, and the encoded picture is multiplexed. The transfer to the device 306 is stopped (step S504).

  Finally, a target code amount of a picture to be encoded next time by the encoding apparatus is set (step S505).

  The target code amount to be set at this time is the prediction after the code amount generated until the buffer failure occurs in the encoding device, and the picture data sufficiently smaller than the code amount is repeatedly output in the other encoding device. Find and set the buffer position.

  By recalculating the target code amount according to the above steps, the buffer occupancy derived from the previous picture generation code amount being encoded by the other encoding device set during normal operation is not used, so continuous buffer failure Avoid in advance.

  The processing flow at the time of buffer failure transmission is as described above.

<Failure reception flow>
Next, a process at the time of failure reception realized by using the above-described functional block, which is a feature of the present invention, will be described with reference to a flowchart of FIG.

  In step S506, reception processing for a buffer failure issued by the other encoding device in step S503 is performed.

  If there is no buffer failure reception (FALSE in step S506), the operation flow is terminated and the normal operation encoding process is continued.

  On the other hand, when a buffer failure is received (TRUE in step S506), the process waits until the current encoding picture is completely encoded (step S507).

  After completion of picture encoding (TURE in step S507), the buffer occupancy is updated from the generated code amount of the picture, and the allocation of the target code amount of the picture to be encoded next is calculated (step S508).

  The target code amount is not a buffer occupancy amount after occurrence of a failed picture, but a sufficiently small target that does not cause a buffer failure by using the buffer occupancy amount at the time when the previous picture encoding is completed in the encoding device. Set the code amount.

  For example, a target code that predicts the buffer occupancy at the decoding time of the picture to be encoded this time from the accumulated amount of data calculated from the current buffer occupancy, the encoding bit rate, and the frame period, and fits in half of the occupancy A method of allocating the amount is effective.

  Then, the target code amount and various encoding parameters are set in the encoder 403, and the picture encoding process is started (step S509).

  Subsequently, the process waits for the completion of coding of the picture for avoiding the failure for which the parameter setting for suppressing the generated code amount is set (step S510), and when the picture coding is completed (TRUE of step S510), repeat output control is performed. A signal for permitting the repetition control of the picture is issued to the unit 401.

  According to the control signal, the encoded picture data is output twice repeatedly at the bitstream output unit 405 (step S511).

  The above is the processing flow in the encoding apparatus at the time of buffer failure reception according to the present invention.

<Output data structure>
Next, FIG. 6 shows the operation timing (FIG. 6 (a)) of each encoding apparatus to which the above-described operation flow according to the present invention is applied, and the data structure of the output bit storm (FIG. 6 (b)).

  The operation timing examples are in accordance with the above-described operation flow, using the two encoding devices of the encoder 1 and the encoder 2, and the order of Pic1, 3, 5, 7, 9 and Pic2, 4, 6, 6, respectively. This is a case where picture coding is performed in the order of 8.

  At this time, it is assumed that an underflow of the buffer has occurred at the time of picture encoding corresponding to Pic3 (601) indicated by shading in the drawing.

  Immediately after the occurrence of the underflow, the encoding in the encoder 1 is immediately terminated, and the occurrence of the buffer failure is transmitted to the encoder 2 at the timing indicated by the arrow 602 in the figure.

  Then, the encoder 2 receives the buffer failure, waits for the completion of the encoding of Pic2, and then, for the subsequent picture Pic4, the code amount corresponding to the period 603 in the figure that occurs from the failure reception to the completion of the encoding Is suppressed based on steps S507 and S508.

  The picture data encoded according to the operation timing is indicated by (b) in the figure by the output control in the bit stream output unit 405.

  In this figure, the output of the buffer failure picture Pic3 is skipped, and instead, the picture Pic4 in which the generated code amount is suppressed and the buffer failure is avoided is output twice so as to have the structure of 604 and 605 in the figure. is there.

  The above is the operation timing of the encoding apparatus according to the present invention and the data structure example of the output bit stream.

<Buffer transition diagram>
Next, FIG. 7 shows a buffer transition of a bit storm output by applying the above-described operation flow and operation example according to the present invention.

  In this figure, the bold solid line is the picture code amount encoded by the encoder 1, and the broken line is the picture code amount encoded by the encoder 2.

  The timing at which the buffer failure avoidance picture Pic4 encoded by the encoder 2 in FIG. 6 and output twice as a bitstream is decoded is the timing of dt1 and dt2 (interval 701 in the figure).

  At this time, as the buffer transition, the picture sweep from the decoding buffer having the same small picture code amount is repeated twice, so that the buffer occupancy amount is underflow and a sufficient free space can be secured at the next picture encoding. become.

  As a result, continuous buffer failure can be avoided and propagation of image quality degradation to the subsequent encoded picture can be prevented to a minimum.

  The above is the buffer transition example of the bit stream output by the encoding apparatus according to the present invention.

(Embodiment 2)
In the first embodiment, when a buffer failure is received, a method has been described in which encoding is performed while suppressing the amount of code when a subsequent picture is encoded, and the encoded picture data is repeatedly output as a bitstream.

  In the second embodiment, when the previous picture is being encoded as the input picture order when the buffer failure is received, the target code amount of picture encoding after the failure reception time is updated, and the encoded picture Is repeatedly output as a bitstream.

  On the other hand, when the subsequent picture is being encoded, the generated code amount after the failed reception time is similarly updated, and the subsequent picture is repeatedly output as a bit stream as a further different operation.

  As a result, if the buffer failure reception is found immediately after the start of encoding, the code amount can be immediately suppressed from the middle of the previous picture without suppressing the generated code amount for the entire subsequent picture screen. Therefore, it is possible to reduce the encoding deterioration range as the output screen.

  Hereinafter, Embodiment 2 of the present invention will be described with reference to the drawings.

  In the description of the second embodiment, portions common to the above-described first embodiment are denoted by the same reference numerals, and description thereof is omitted as appropriate.

<Block configuration>
FIG. 8 shows a block configuration of a video encoding apparatus to which the second embodiment can be applied.

  The encoding apparatus includes an encoded picture order determination unit 801 in addition to the blocks of the first embodiment.

  When the encoded picture order determining unit 801 receives a buffer failure, the encoded picture order determining unit 801 determines the encoding order of the buffer-failed picture and the picture being encoded by the encoded picture unit 403.

  Based on the result of the determination unit, when the picture before the failed picture is coded, the generated code amount of the previous picture is suppressed and the repeat output is controlled. When the subsequent picture is generated, the generated code amount of the subsequent picture is suppressed and the repeat output is controlled. Switch to control.

  Other block functions are the same as those in the first embodiment, and thus the description thereof is omitted.

<Failure reception flow>
FIG. 9 shows an operation flow at the time of failure reception realized using the above-described functional blocks.

  Note that the operation flow at the time of buffer failure transmission is the same as that of the first embodiment, and is therefore omitted.

  First, in step S506, reception processing for a buffer failure issued by the other encoding device is performed.

  If no buffer failure has been received (FALSE in step S506), the operation flow is terminated and the normal operation encoding process is continued.

  On the other hand, if a buffer failure is received (TRUE in step S506), the encoding operation of the picture encoder 403 is temporarily stopped, and the generated code amount that has already been encoded is acquired (step S901).

  Subsequently, based on the encoded picture data size acquired in step S901, the allocation of the target code amount is recalculated so as not to underflow the buffer for the remaining generated code amount of the encoded picture (step S902).

  The target code amount does not use the buffer occupancy amount after occurrence of the failed picture, but the current occupancy amount from the sum of the buffer occupancy amount at the start of encoding of the encoded picture and the encoded picture data acquired in step S901. And a sufficiently small target code amount that does not cause buffer failure is set.

  For example, a target code amount that predicts the buffer occupancy amount in the decoding time of the encoded picture from the accumulated data amount calculated from the current buffer occupancy amount, the encoding bit rate, and the frame period, and falls within half of the occupancy amount A method of assigning the value is effective.

  Then, the target code amount and various encoding parameters are set in the encoder 403, and the encoding process of the picture is restarted (step S903).

  Thereafter, the processing in steps S510 and S511 is executed in the same manner as in the first embodiment, and this operation flow is terminated.

  The above is the processing flow at the time of buffer failure reception.

<Output data structure>
Subsequently, FIG. 10 shows an operation timing chart of each encoding apparatus to which the above-described operation flow according to the present invention is applied and a data structure of an output bit storm.

  FIG. 10A shows a case where a buffer failure at the time of Pic3 picture encoding in the encoder 1 occurs during Pic4 encoding of the encoder 2, and a case where a buffer failure is received in the frame period of t2. It is.

  On the other hand, FIG. 10B shows a case where a buffer failure at the time of Pic3 picture encoding in the encoder 1 occurs during Pic2 encoding of the encoder 2, and the buffer failure is received in the frame period of t1. This is the case.

  As described above, the bitstream output when this operation flow is executed varies depending on the coding order of the coded pictures that are processed in parallel by the encoder 2 when the buffer failure occurs.

  During Pic4 encoding, the encoded picture Pic4 is repeated twice in the same manner as in FIG.

  When Pic2 encoding is in progress, the data structure is such that the generated code amount of the encoded picture Pic2 is suppressed and the bitstream is output repeatedly.

  In the first embodiment, the subsequent picture Pic4 that follows the failed picture is repeatedly output with code amount suppression regardless of the occurrence timing of the buffer failure. However, feedback is immediately performed even during picture coding during parallel operation. changed.

  The above is the operation timing of the encoding apparatus according to the present invention and the data structure example of the output bit stream.

<Buffer transition diagram>
Next, FIG. 11 shows a bit storm buffer transition output by applying the operation flow according to the present invention.

  FIG. 11A shows the buffer transition when the code amount of Pic4 following the failed picture Pic3 is suppressed and the bitstream is repeatedly output, as in the first embodiment. At the timing of dt0, Pic2 is swept from the buffer, and the picture data Pic4 with the code amount suppressed at the timing of dt1 and dt2 is decoded and output.

  On the other hand, FIG. 11B shows buffer transition when the generated code amount is suppressed for Pic2 before the failed picture Pic3 and the bitstream is repeatedly output.

  The difference from FIG. 11A is a transition in which the same picture data Pic2 is repeatedly swept from the buffer at the timing of dt0 and dt1, and the picture Pic4 is decoded and output at the timing of dt2.

  In both cases, as in the first embodiment, continuous buffer underflow is avoided, and the code amount is reset for Pic2 in the middle of encoding as shown in FIG. It is possible to minimize image quality degradation that extends to.

  The above is the buffer transition example of the bit stream output by the encoding apparatus according to the present invention.

(Embodiment 3)
In the second embodiment, the order of encoded pictures at the time of receiving a buffer failure is determined, and based on the determination result, the target code amount of either the current encoded picture or the subsequent picture is updated, and the code A method of repeatedly outputting a bitstream of a picture has been described.

  In the third embodiment, in addition to the second configuration, when a buffer failure is received during previous picture encoding, the screen difficulty level of the encoding target image corresponding to the subsequent picture stored in advance in the frame buffer memory Is compared with the screen difficulty level of the picture data currently being encoded.

  Then, based on the comparison result, selecting a picture image with a low screen difficulty level and applying the generated code amount suppression is performed as a different operation.

  As a result, even if the generated code amount is suppressed and the picture data is repeated, it is possible to achieve failure avoidance using an image that is less prominent in image quality degradation during reproduction.

  Note that the operation according to the present embodiment is applicable only when the encoding order of the picture being encoded is encoding a picture preceding the corrupted picture.

  Hereinafter, Embodiment 3 of the present invention will be described with reference to the drawings.

  In the description of the third embodiment, the same reference numerals are given to portions common to the above-described first and second embodiments, and description thereof is omitted as appropriate.

  FIG. 12 shows a block configuration of a video encoding apparatus to which the third embodiment can be applied.

  This moving image coding apparatus is characterized by comprising a screen difficulty level determination unit 1201 in addition to the blocks of the second embodiment.

  The screen difficulty level determination unit 1201 reads the current image data to be encoded next time stored in advance in the buffer memory in the encoding device, calculates the screen difficulty level of the image prior to encoding, and the encoding device The comparison is made with the screen difficulty level of the picture data currently being encoded.

  Therefore, in the configuration of the third embodiment, an area of the frame buffer memory 302 that stores the image input from the distributor 301 is secured for a size of two screens or more.

  As a method for obtaining the screen difficulty level performed by the determination unit 1210, for example, a method for deriving an activity value for a target image in code amount control employed in the MPEG-2 software model “TM5 (TestModel)” is effective. is there.

  For the activity value, the magnitude of the change in the luminance value in the block is obtained in units of 4 × 4 or 8 × 8 pixel blocks. Therefore, since it can be estimated whether or not the image has a lot of high-frequency components, it can be determined whether or not the image is easily noticeable in encoding degradation.

  That is, it can be determined that the smaller the activity value is, the more uniform and flat the pattern in the screen is.

  In general, if the suppression of the amount of generated code is increased, the quantized value of the image data frequency-converted using orthogonal transform such as DCT (Discrete Cosine Transform) to the high-frequency component increases, so that the original image has a high frequency component Deterioration is conspicuous.

  In the embodiment of the present invention, the activity value is obtained for the picture image encoded at the time of buffer failure reception and the picture image scheduled to be encoded next time, and the two activity values are compared.

  For the above reason, the code amount is suppressed for a picture having a small activity value, and deterioration during reproduction is made inconspicuous.

  Other block functions are the same as those in the first embodiment, and thus the description thereof is omitted.

<Failure reception flow>
FIG. 13 shows an operation flow at the time of failure reception realized using the above-described functional blocks.

  Note that the operation flow at the time of buffer failure transmission is the same as in the first and second embodiments, and is omitted.

  First, in step S506, reception processing for a buffer failure issued by the other encoding device is performed.

  If no buffer failure has been received (FALSE in step S506), the operation flow is terminated and the normal operation encoding process is continued.

  On the other hand, if a buffer failure has been received (TRUE in step S506), the encoded picture determination unit 801 determines whether the picture currently being encoded is before or after the broken picture as a reference (step S506). S1201).

  When the subsequent picture is being encoded (FALSE in step S1201), the processing in steps S901, S902, S903, S510, and S511 is performed in the same manner as in the second embodiment to suppress the code amount of the subsequent picture and output the repeated bitstream. This operation flow ends.

  If the previous picture is being encoded (TRUE in step S1201), the frame buffer memory 303 in the encoder 302 is accessed, and the screen difficulty level of the current image to be encoded next time that has already been stored. Is calculated and acquired (step S1202).

  Then, in the screen difficulty level determination unit 1201 block, the screen difficulty level acquired in step S1202 and the screen difficulty level of the currently-encoded picture are compared, which is less complex in the screen (step S1203).

  From the comparison result in step S1203, it is determined whether the picture to be encoded next time acquired in step S1202 has a smaller complexity in the screen (step S1204).

  If the picture currently being encoded is less difficult than the picture to be encoded next time (FALSE in step S1204), the process proceeds to step S901, and the current encoding is performed according to the same flow as in the second embodiment. This operation flow is finished by repeatedly suppressing the code amount of the middle picture and outputting the picture data stream.

  On the other hand, if the picture to be encoded next time is less difficult (TRUE in step S1204), the process proceeds to step S507, and the code amount suppression of the subsequent picture and the stream of picture data are performed by the same flow as in the first embodiment. Output is repeated, and this operation flow ends.

  The above is the processing flow at the time of buffer failure reception.

  According to the above-described operation flow, the data structure and buffer transition of the output bit stream are the same as those described with reference to FIGS.

  That is, when the next encoded image is less difficult than the picture being encoded, the code amount of the succeeding picture is suppressed, and therefore (a) in FIG. 10 and (a) in FIG. And the same data structure and buffer transition.

  If the picture being coded has a lower screen difficulty level, code amount suppression and repeated bitstream output are performed for the picture currently being coded (the picture before the failed picture). The data structure and buffer transition are the same as in (b) and FIG.

  However, as a difference from the case of the second embodiment, since the target picture for which the generated code amount is to be suppressed is selected by the screen difficulty level determination, it is possible to realize an effect that the image quality deterioration of the playback screen of the output bitstream is less noticeable.

(Embodiment 4)
In the first to third embodiments, the buffer failure transmission / reception and the failure recovery method in a system configuration in which two encoding devices are processed in parallel have been described.

  In the fourth embodiment, on the premise of a system configuration in the case where three encoding devices are processed in parallel, more general control of the system configuration and the failed transmission / reception method is performed as a different operation.

  As a result, the failure avoidance process can be realized even in a large-scale encoder system using three or more encoding devices.

<System configuration>
FIG. 14 shows a system configuration of an encoding apparatus applied in the fourth embodiment of the present invention.

  The encoding apparatus includes a third encoder 1401 and a buffer failure manager 1402 in addition to the system configuration described in the first embodiment.

  The third encoding device 1401 performs the same operation as in the first embodiment.

  The buffer failure manager 1402 has a function of receiving a buffer failure transmitted from each of the encoders 302, 304, and 1401 and transmitting buffer failure information to each encoder.

  In the embodiments described above, on the premise of the configuration of only two encoding devices, the buffer failure signal is directly transmitted and received between the encoders. However, in the case of a configuration having three or more encoders, It is necessary to avoid that the failed transmission / reception loops to the same encoder and that there are multiple encoding apparatuses that repeatedly output picture data.

  Therefore, the present embodiment is characterized in that the buffer failure manager 1402 is used to centrally manage which encoding device has a buffer failure.

  As an operation of the failure manager 1402, for example, when a buffer failure occurrence is received by the encoder 302, a buffer failure transmission is transmitted only to either the second encoder 304 or the third encoder 1401. Has a good mask function.

  Therefore, in the buffer failure transmission / reception information (1403, 1404) connecting each encoder and the failure manager 1402, in addition to the presence / absence of failure, an identifier uniquely assigned in advance in this system is identified. Information shall be transmitted.

  The above is the system configuration in Embodiment 4 of the present invention.

<Operation Example of Embodiment 4>
Next, the operation timing of the failure recovery process and the data structure example when a buffer failure picture occurs in the system configuration will be described with reference to FIGS. 15 and 16.

  Note that the processing block in each encoder and the operation flow at the time of failed transmission / reception are the same as those in the first to third embodiments, and will be omitted.

  FIG. 15 is an operation timing when three encoders are processed in parallel, and is an example in which each encoder operates in a pipeline while allocating three times the frame period of the input video to the encoding process. .

  In the encoder 1 shown in FIG. 15, when a buffer failure occurs during picture encoding corresponding to Pic4 (1501) filled with shading, the failure information is once sent from the encoder 1 to the buffer failure manager 1402 described above. And the buffer failure is transmitted again only to the encoder 2 or the encoder 3.

  Then, when a buffer failure is transmitted to the encoder 2, the suppression of the generated code amount of the picture after the picture where the buffer failure has occurred, Pic5 (1502), is transmitted to the encoder 3. In this case, the generated code amount of Pic6 (1503) is suppressed according to the buffer failure reception operation flow up to the first to third embodiments.

  The bit stream output after performing the encoding process at the above timing has a data structure as shown in FIG. 16 by the encoder that has received the buffer failure.

  In this figure, the picture data Pic5 from the encoder 2 is output twice (FIG. 16 (a)), and the picture data Pic6 from the encoder 3 is output twice (FIG. 16 (b)). ) Respectively.

  In the embodiment of the present invention, it is not particularly limited to which encoder other than the encoder in which the failure has occurred, and the failure occurrence is transmitted. However, a transmission path predetermined as an encoding system may be used.

  For example, if the encoder 1 fails, the data is transmitted to the encoder 2. If the encoder 2 fails, the data is transmitted to the encoder 3. If the encoder 3 fails, the data is transmitted to the encoder 1. If it is a combination such as this, the logic of the buffer failure manager can be simplified.

  Although the present embodiment has been described in a system configuration using three encoding devices, it can be applied universally to a parallel operation system using four or more encoding devices.

301: Distributor 302: Encoder 303: Buffer memory 304: Second encoder 305: Buffer failure communication port 306: Multiplexer 401: Repeat output control unit 402: Buffer failure receiver 403: Picture encoder 404: Target code amount setting unit 405: Bit stream output unit 406: Buffer failure transmission unit

Claims (3)

One input video signal is distributed to multiple encoders, and each encoder performs in-picture encoding processing in units of pictures in parallel, and then outputs a single bitstream multiplexed with multiple bitstreams. A moving image encoding device,
Buffer failure detection means for detecting a buffer failure during the encoding process by the encoder;
Buffer failure transmitting means for transmitting information on occurrence of buffer failure to the other encoder;
Buffer failure receiving means for receiving information on occurrence of the buffer failure; and
Coding control means for performing coding while suppressing the generated code amount after occurrence of the buffer failure with respect to the picture next to the coded picture;
Picture data discarding means for discarding a bitstream output of encoded picture data in which a buffer failure has occurred;
Picture data repeat output means for outputting the next picture of the encoded picture as a repeated bit stream;
A moving picture encoding apparatus comprising: The encoding control means suppresses the remaining generated code amount in the encoded picture, and encoding pause means for temporarily stopping the encoding process of the encoded picture when the buffer failure occurrence information is received. The moving picture encoding apparatus according to claim 1, further comprising encoding restarting means for restarting encoding. The encoding control unit suppresses the amount of generated code for a picture with a lower difficulty level based on the comparison by the comparison unit comparing the encoding difficulty level between the encoded picture and the subsequent picture, and the comparison unit. 3. The moving picture encoding apparatus according to claim 1, further comprising: a picture selection unit that switches to be the target of the video.