JP6311821B2 - Moving image processing apparatus and moving image processing method - Google Patents

Description

  The present invention relates to a moving image processing apparatus and a moving image processing method for dividing each picture into a plurality of blocks and processing each block.

  In recent video coding, a high compression rate is achieved by dividing an image into blocks, predicting pixels included in the block, and coding a prediction difference. A prediction mode that configures a prediction pixel from pixels in a picture to be encoded is called intra prediction, and a prediction mode that configures a prediction pixel from a previously encoded reference image called motion compensation is called inter prediction.

  In the inter-prediction apparatus, in inter prediction, a region referred to as a prediction pixel is expressed by two-dimensional coordinate data of horizontal and vertical components called motion vectors, and motion vector and pixel prediction difference data are encoded. In order to suppress the coding amount of the motion vector, a prediction vector is generated from a motion vector of a block adjacent to the encoding target block, and a difference vector between the motion vector and the prediction vector is encoded.

  MPEG (Moving Picture Experts Group) -4 AVC / H.M. In H.264 (hereinafter also referred to as H.264), a moving picture standard represented by HEVC (High Efficiency Video Coding) under normalization, each block is addressed in a raster order, and the processing order is the address of the address. Specified in order.

  In the moving image encoding / decoding process, the number of pixels processed per second is enormous, and high calculation performance is particularly required for motion compensation and orthogonal transformation. Therefore, parallel processing is important. In moving picture coding, there is a dependency relationship between blocks in order to calculate various prediction values from peripheral blocks that have been processed like the above-described prediction vectors. Therefore, it is difficult to perform parallel processing for each block.

  As a technique for achieving parallel processing for each block without affecting the processing order and block dependency on the standard, there is a method for performing parallel processing for each block line while shifting the horizontal position of the processing block.

  Hereinafter, such parallel processing will be referred to as block line parallel processing. Next, the case of performing block line parallel processing of two block lines will be briefly described.

  The moving image processing apparatus has processing units for processing block lines in order to process block lines in parallel. For example, unit 1 processes odd block lines and unit 2 processes even block lines.

  FIG. 1 is a diagram for explaining block line parallel processing. As shown in FIG. 1, the horizontal address of the block of the first block line (first block line) processed by the unit 1 and the horizontal address of the block of the second block line (second block line) processed by the unit 2 are processed. Shift the address by 2 blocks or more.

  Accordingly, when attention is paid to the processing block X of the unit 2, the upper left block B, the upper block C, and the upper right block D processed by the unit 1 have been processed together with the left block A of the processing block X. In block X, the encoded results of these processed blocks can be used.

  For example, since processing of even block lines can be started without waiting for the end of odd block lines, parallelization can be achieved with respect to processing such as motion prediction and orthogonal transformation. In the above, the case of parallel processing of two block lines has been described. However, in the case of parallel processing of N block lines, N units may be allocated to the moving image processing apparatus in units of N block lines.

  When executing block line parallel processing as a program, the processing unit may be a thread or a CPU (Central Processing Unit).

  H. In H.264, with respect to entropy coding, the order of output bit sequences is processed in the order of processing blocks according to the standard. Therefore, the moving image processing apparatus temporarily stores the result of the above-described block line parallel processing, and performs entropy encoding processing in the order of processing blocks according to the standard.

  On the other hand, HEVC discloses a technique for interleaving the order of output bit sequences for each block line. FIG. 2 is a diagram for explaining block line parallel processing of an entropy processing unit in HEVC. As shown in FIG. 2, parallel processing can be performed for each block line also for entropy encoding / decoding processing.

ISO / IEC 14496-10 (MPEG-4 Part 10) / ITU-T Rec.H.264 Thomas Wiegand, Woo-Jin Han, Benjamin Bross, Jens-Rainer Ohm, Gary J. Sullivan, "Working Draft 5 of High-Efficiency Video Coding" JCTVC-G1103, JCT-VC 7th Meeting, December 2011 HEVC reference software HM5.0 "https://hevc.hhi.fraunhofer.de/svn/svn_HEVCSoftware/tags/HM-5.0/" MPEG-2, Test Model 5 (TM5), Doc.ISO / IEC JTC1 / SC29 / WG11 / N0400, Test Model Editing Committee, April 1993.

  H. In a hybrid coding method that combines motion compensation typified by H.264 and HEVC and orthogonal transform processing (for example, DCT (Discrete Cosine Transform)), an image is divided into blocks, and predicted pixels of each block are generated, and original pixels are generated. The compression is realized by performing orthogonal transformation such as DCT on the difference pixel between the prediction pixel and the prediction pixel, and quantizing the output coefficient of the orthogonal transformation.

  In order to control the amount of information, a quantization parameter (also called a QP (Quantization Parameter)) for adjusting the accuracy of quantization is prepared, and a QP value is encoded for each block.

  However, when the transform coefficient is a block of all 0, the result of inverse quantization is also all 0, and no QP is required in the decoding process. In such a case, this QP is invalid and the QP value is encoded. Not.

  Since a predicted value of QP (also referred to as a QP predicted value) is generated in each block, when the QP of the processing block becomes invalid, the QP value of the processing block is set to the predicted value of QP. As a method for determining a QP value for each block, an algorithm used in TM5 described in Non-Patent Document 4 is known.

H. The encoding method of the QP value in H.264 or HEVC encodes the difference value QP_DELTA between the QP value of the processing block and the QP prediction value. The QP prediction value is, for example, QPprev that is a QP value immediately before the processing block in the raster order. QP_DELTA is calculated by the following equation (1).
QP_DELTA = QP−QPprev (1)
The moving image decoding apparatus decodes the QP_DELTA that has been entropy-coded in the moving image encoding apparatus, and restores the QP value in Equation (2).
QP = QP_DELTA + QPprev (2)
If there is no previous block such as the first block of the processed picture, QPprev may be a value that can be determined prior to processing of the first block. For example, H.M. In H.264, the QPprev of the picture head block is the SliceQP value described in the header information called Slice.

  H. In H.264 and HEVC, in addition to orthogonal transform quantization, the QP value of each block is used to determine the filter strength of the deblocking filter. Since the invalidated QP is not notified to the video decoding device, the QP value of the block in which the QP is invalidated is processed as QPprev.

  Here, attention is focused on the K-1 block line and the K block line in the block line parallel processing. When the first X block of the K block line is processed at the start of the processing of the K block line, the processing is completed only to the extent that the K-1 block line is shifted by two blocks. Since the block located before in the raster order of the first block of the K block line is the last block Y of the K-1 block line, the processing of the Y block has not been completed yet.

  For this reason, when processing an X block, there exists a subject that QPprev is not fixed. Since QPprev of the X block is not fixed, QPprev of the next block after the X block is also not fixed.

  Therefore, the calculation of the QP prediction value cannot be performed in parallel for each block line, and is a sequential process. FIG. 3 is a block diagram showing a schematic configuration of a moving image processing apparatus that performs conventional block line parallel processing.

  FIG. 3A is a block diagram showing a schematic configuration of a conventional encoding apparatus that performs block line parallel processing. FIG. 3B is a block diagram illustrating a schematic configuration of a conventional decoding apparatus that performs block line parallel processing.

  As shown in FIG. 3, in the conventional moving image processing apparatus, a QP prediction unit that calculates a QP prediction value becomes a bottleneck.

  FIG. 4A is a diagram illustrating block line parallel processing (part 1) in the QP prediction processing. In the example shown in FIG. 4A, it is assumed that the first block line and the second block line are processed in parallel. At this time, when the processing block of the second block line is X, it is necessary to wait until the processing of the block Y is completed in order to calculate the QP predicted value of X. The processing of the second block line cannot be started without completing the processing of the first block line.

  FIG. 4B is a diagram illustrating block line parallel processing (part 2) in the QP prediction processing. As shown in FIG. 4B, when the processing of the block Y is completed, the processing of the second block line can be started.

  FIG. 4C is a diagram illustrating block line parallel processing (part 3) in the QP prediction processing. In the example shown in FIG. 4C, in order for the QP prediction unit that has processed the first block line to calculate the QP value of the block W of the third block line, the processing of the block V of the second block line is performed. Wait until is completed.

  That is, the processing of the third block line cannot be started unless the processing of the second block line is completed. As described above, in order to start the processing of the Kth block line, it is required that the processing of the (K-1) th block line is completed. It becomes sequential processing.

  When the encoding processing unit is parallelized and the QP prediction unit is sequentially processed, the following problem occurs. Since QPprev in block X has not been determined, for example, QP_DELTA cannot be calculated at the head of the Kth block line, and the problem that entropy processing of block X cannot be started until processing of block Y is completed occurs. .

  Therefore, since the start of the entropy process is delayed, the number of buffers for storing the results of the encoding processing unit increases. Similarly, there also arises a problem that the deblocking filter process cannot be started.

  Similarly, block line parallel processing is considered for the moving picture decoding apparatus. In HEVC, the entropy decoding process can be performed in parallel for each block line. Similarly to the above, paying attention to the K-1th block line and the Kth block line, and considering the QP restoration processing of the first block X of the Kth block line, when the QP_DELTA of the block X can be decoded, Since QP_DELTA of block Y has not been decoded, the QP value of Y cannot be restored.

  Since QPprev that is the QP predicted value of block X is the QP value of block Y, the QP predicted value of block X cannot be calculated, and the QP value of block X cannot be restored. For this reason, the calculation of the QP prediction value of each block is a sequential process as in the case of the moving picture encoding apparatus, which becomes a bottleneck of parallel processing.

  Therefore, the disclosed technique aims to make it possible to parallelize the calculation processing of the QP prediction value and to improve the efficiency of the block line parallel processing.

A moving image processing apparatus according to an aspect of the disclosure is provided.
A moving image processing apparatus including a moving image encoding unit that divides an image into a plurality of blocks and performs a moving image encoding process for each block, and a moving image decoding unit that decodes an encoded stream. ,
The moving image encoding unit includes:
A parallel encoding processing unit that generates, in parallel for each block line, encoded data including quantized orthogonal transform coefficients for blocks included in the block line indicating the row of blocks;
A first parallel QP prediction unit that calculates a prediction value of a quantization parameter for a quantization parameter used in encoding of each block included in the block line in parallel for each block line;
A parallel entropy encoding unit that performs entropy encoding on each block included in the block line using the encoded data and the predicted value of the quantization parameter in parallel for each block line, and
The first parallel QP prediction unit includes:
When parallel processing is performed in units of N block lines each having a value of 2 or more, the horizontal position of the processing block of the (K-1) th block line is at least one block ahead of the horizontal position of the processing block of the Kth block line. To process
When the processing block is at the head of the block line, the quantization parameter of the block processed immediately before the processing block is initialized with the quantization parameter determined in units of the slice to which the processing block belongs, and the initialized quantization parameter is set. A predicted value of the quantization parameter of the processing block based on
In the case where the processing block is other than the head of the block line, the predicted value of the quantization parameter of the processing block is calculated based on the quantization parameter of the block in the block line,
The parallel encoding processing unit
The quantization parameter of the processing block based on the prediction value of the quantization parameter calculated by the first parallel QP prediction unit, the quantization parameter used for the processing of each block, and whether the result of encoding is all 0 or not Decide
The moving image decoding unit
A parallel entropy decoding unit that entropy-decodes a stream of blocks included in a block line indicating the row of blocks in parallel for each block line;
A second parallel QP prediction unit that calculates a prediction value of a quantization parameter of each block included in the block line in parallel for each block line;
For each block included in the block line, a decoded pixel obtained by performing decoding processing using the data decoded by the parallel entropy decoding unit and the prediction value calculated by the second parallel QP prediction unit, A parallel decoding processing unit that generates each in parallel,
The second parallel QP prediction unit includes:
When parallel processing is performed for each of N block lines indicating a value of 2 or more, the relationship between the horizontal position of the processing block of the (K-1) th block line and the horizontal position of the processing block of the Kth block line is K-1 The timing at which the horizontal position of the processing block of the first block line precedes at least one block,
When the processing block is at the head of the block line, the quantization parameter of the block processed immediately before the processing block is initialized with the quantization parameter determined in units of the slice to which the processing block belongs, and the initialized quantization parameter is set. A predicted value of the quantization parameter of the processing block based on
In the case where the processing block is other than the head of the block line, the prediction value of the quantization parameter of the processing block is calculated based on the quantization parameter of the block that has been decoded earlier in the block line,
The parallel decoding processing unit
The quantization parameter of the processing block is calculated based on the predicted value of the quantization parameter calculated by the second parallel QP prediction unit, and the deblocking filter process of the processing block is performed using the calculated quantization parameter.

  According to the disclosed technique, it is possible to parallelize the calculation processing of the QP prediction value, and to improve the efficiency of the block line parallel processing.

The figure for demonstrating block line parallel processing. The figure for demonstrating the block line parallel processing of the entropy process part in HEVC. The block diagram which shows schematic structure of the encoding apparatus which performs the conventional block line parallel processing. The block diagram which shows schematic structure of the decoding apparatus which performs the conventional block line parallel processing. The figure which shows the block line parallel process (the 1) in QP prediction process. The figure which shows the block line parallel process (the 2) in QP prediction process. The figure which shows the block line parallel process (the 3) in QP prediction process. 1 is a block diagram illustrating an example of a configuration of a video decoding device in Embodiment 1. FIG. FIG. 3 is a block diagram illustrating an example of a configuration of each unit of the parallel decoding unit according to the first embodiment. FIG. 3 is a block diagram illustrating an example of a configuration of a decoding processing unit according to the first embodiment. The figure which shows an example of the block which can be utilized in the parallel QP prediction process in case of N = 3. FIG. 3 is a block diagram illustrating an example of a configuration of a QP prediction unit according to the first embodiment. The figure which shows an example of the QP prediction process (the 1) in Example 1. FIG. The figure which shows an example of the QP prediction process (the 2) in Example 1. FIG. 3 is a flowchart illustrating an example of a block decoding process according to the first embodiment. FIG. 9 is a block diagram illustrating an example of a configuration of a moving image encoding device according to a second embodiment. FIG. 9 is a block diagram illustrating an example of a configuration of each unit of a parallel encoding unit according to the second embodiment. FIG. 9 is a block diagram illustrating an example of a configuration of an encoding processing unit according to the second embodiment. FIG. 10 is a block diagram illustrating an example of a configuration of a QP prediction unit according to a second embodiment. 10 is a flowchart illustrating an example of a block encoding process according to the second embodiment. FIG. 9 is a block diagram illustrating an example of a configuration of a moving image processing apparatus according to a third embodiment.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. Note that the image (picture) included in the moving image may be either a frame or a field. The frame is one still image in the moving image data, while the field is a still image obtained by extracting only odd-numbered data or even-numbered data from the frame.

  Further, the moving image to be processed may be a color moving image or a monochrome moving image.

[Example 1]
<Configuration>
FIG. 5 is a block diagram illustrating an example of the configuration of the video decoding device according to the first embodiment. In the example illustrated in FIG. 5, the video decoding device 10 includes a parallel decoding unit 101 and a decoded pixel storage unit 105. The parallel decoding unit 101 includes a parallel entropy decoding unit 102, a parallel QP prediction unit 103, and a parallel decoding processing unit 104.

  Note that the video decoding device 10 performs parallel processing on N block lines. Further, the block line refers to a horizontal block row in the image, for example.

  Each unit of the parallel decoding unit 101 sets the same block to be processed in each block line. Further, for example, the horizontal position of the processing block of the (K-1) th block line is two blocks or more ahead of the horizontal position of the processing block of the Kth block line.

  This is because the horizontal position is shifted by two or more blocks between block lines, so that the decoding information of the upper and upper right blocks can be used even if each block line is decoded in parallel. Because. When using decoding information of only the upper block, the horizontal position shift amount between the block lines may be one block. In the following, the upper block indicates a block on one block of the processing block, and the upper right block indicates a block adjacent to the right of the block on one block of the processing block.

  The parallel decoding unit 101 decodes an image divided into a plurality of blocks by performing parallel processing on, for example, N block lines on a stream encoded using a moving image encoding method. Pixels decoded for each block of the block line are output to the decoded pixel storage unit 105. The decoded pixel is referred to as a decoded pixel.

  The parallel entropy decoding unit 102 divides the input stream for each block line, and performs entropy decoding of the block stream included in the block line in parallel for each block line. The parallel entropy decoding unit 102 outputs the entropy decoded data to the parallel decoding processing unit 104.

  The parallel QP prediction unit 103 calculates the prediction value (QP prediction value) of the quantization parameter (QP) of each block included in the block line in parallel for each block line. The calculated QP prediction value is output to the parallel decoding processing unit 104.

  The parallel decoding processing unit 104 decodes each block included in the block line using the data decoded by the parallel entropy decoding unit 102 and the QP prediction value calculated by the parallel QP prediction unit 103. Are generated in parallel for each block line. The generated decoded pixel is output to the decoded pixel storage unit 105.

  The decoded pixel storage unit 105 stores the decoded pixel of each block output from the parallel decoding unit 101. Decoded pixels form an image (picture) when aggregated in units of images. The decoded pixel storage unit 105 outputs a decoded image when the decoding process for one picture is completed.

《Parallel decoding unit》
Next, the parallel decoding unit 101 will be described. FIG. 6 is a block diagram illustrating an example of a configuration of each unit of the parallel decoding unit 101 according to the first embodiment. In the example illustrated in FIG. 6, the parallel entropy decoding unit 102 includes a first entropy decoding unit 221, a second entropy decoding unit 222, and an Nth entropy decoding unit 223.

  In the example illustrated in FIG. 6, the parallel QP prediction unit 103 includes a first QP prediction unit 231, a second QP prediction unit 232, and an NQP prediction unit 233. In the example illustrated in FIG. 6, the parallel decoding processing unit 104 includes a first decoding processing unit 241, a second decoding processing unit 242, an Nth decoding processing unit 243, and a block information holding unit 244.

  Here, for L = 1 to N, the Lth entropy decoding unit, the LQP prediction unit, and the Lth decoding processing unit process one block line correspondingly. Hereinafter, the L-th entropy decoding unit, the LQP prediction unit, and the L-th decoding processing unit are collectively referred to as a block line decoding processing unit. For example, the block line decoding processing unit 201 includes a first entropy decoding unit 221, a first QP prediction unit 231, and a first decoding processing unit 241.

  When the block line decoding processing unit processes the Kth block line, it next processes the K + Nth block line.

  The parallel entropy decoding unit 102 divides the input stream for each block line. Each entropy decoding unit 221 to 223 performs entropy decoding of each block line in parallel in units of N block lines in the divided stream. Each entropy decoding part 221-223 performs the entropy decoding process corresponding to the entropy encoding process in a moving image encoder.

  Each of the decoding processing units 241 to 243 performs decoding processing in parallel on a block line basis. FIG. 7 is a block diagram illustrating an example of the configuration of the decoding processing unit according to the first embodiment. Since each of the decoding processing units 241 to 243 performs the same processing, the following description will be given using the first decoding processing unit 241.

  The first decoding processing unit 241 includes a QP restoration unit 301, an inverse quantization unit 302, an inverse orthogonal transform unit 303, a motion vector restoration unit 304, a prediction pixel generation unit 305, and a decoded pixel generation unit 306. .

  The QP restoration unit 301 restores a QP value using a QP prediction value input from the first QP prediction unit 231 described later and a QP difference value input from the first entropy decoding unit 221 (for example, Expression (2)) reference). The restored QP value is output to the inverse quantization unit 302.

  The inverse quantization unit 302 performs inverse quantization by multiplying the restored QP value and the orthogonal transform coefficient. The inversely quantized orthogonal transform coefficient is output to the inverse orthogonal transform unit 303.

  The inverse orthogonal transform unit 303 performs an inverse orthogonal transform process on the orthogonal transform coefficient input from the inverse quantization unit 302 to generate a prediction error pixel. The generated prediction error pixel is output to the decoded pixel generation unit 306.

  The motion vector restoration unit 304 acquires the motion vector information of the peripheral blocks of the processing block from the block information storage unit 244, and calculates a prediction vector. The motion vector restoring unit 304 restores the motion vector by adding the prediction vector and the difference vector input from the first entropy decoding unit 221. The restored motion vector is output to the predicted pixel generation unit 305.

  The predicted pixel generation unit 305 acquires pixel data of the reference picture indicated by the motion vector from the decoded pixel storage unit 105 in which previously decoded pictures are stored, and generates a predicted pixel. The generated prediction pixel is output to the decoded pixel generation unit 306.

  The decoded pixel generation unit 306 adds the prediction pixel input from the prediction pixel generation unit 305 and the prediction error pixel input from the inverse orthogonal transform unit 303 to generate a decoded pixel. The generated decoded pixel is stored in the decoded pixel storage unit 105.

  Next, processing of each QP prediction unit will be described. First, blocks that can be used in prediction processing by each QP prediction unit will be described. FIG. 8 is a diagram illustrating an example of an available block in the parallel QP prediction process when N = 3. In the example shown in FIG. 8, N = 3, and each block line is delayed in parallel by two blocks.

  At this time, usable blocks for the processing block X are the shaded blocks shown in FIG. The shaded blocks shown in FIG. 8 indicate blocks that have already been processed, that is, blocks that can be used when processing the processing block. The bold-lined blocks shown in FIG. 8 indicate blocks to be processed (also referred to as processing blocks). Each QP prediction unit obtains a QP prediction value by referring to an available block (processing completion block) when calculating the QP prediction value of the processing block. Hereinafter, a method for obtaining the QP predicted value will be described in detail.

  FIG. 9 is a block diagram illustrating an example of the configuration of the QP prediction unit according to the first embodiment. Since each of the QP prediction units 231 to 233 performs the same process, the following description will be given using the first QP prediction unit 231.

  The first QP prediction unit 231 illustrated in FIG. 9 includes a previous QP storage unit 401, a QP selection unit 402, and an upper QP acquisition unit 403.

  The immediately preceding QP storage unit 401 receives the QP value of the block processed immediately before the current processing block from the QP restoration unit 301 and stores it. The QP value stored in the immediately preceding QP storage unit 401 is initialized at the start of picture processing.

  For example, H.M. Similarly to H.264, initialization is performed with the Slice QP value encoded in the Slice header information. Slice is a unit when a block belonging to one picture is divided into groups.

  The upper QP acquisition unit 403 acquires, from the block information storage unit 244, for example, the QP value of the block located above the processing block.

  The QP selection unit 402 selects the QP value output from either the immediately preceding QP storage unit 401 or the upper QP acquisition unit 403 and outputs the selected QP value to the QP restoration unit 301 as a QP predicted value.

  For example, the QP selection unit 402 selects the QP value output from the upper QP acquisition unit 403 if the processing block is the block block head block, and if the processing block is a block other than the block line head block, The QP value output from the storage unit 401 is selected. The QP selection unit 402 outputs the selected QP value to the QP restoration unit 301 as a QP predicted value.

  The prediction (selection) of the QP value in the case of the above example will be described with reference to the drawings. FIG. 10A is a diagram illustrating an example of the QP prediction process (part 1) according to the first embodiment. In the example shown in FIG. 10A, when the position of the processing block X is the processing block and the processing block X is the first block as shown in FIG. 10A, the QP selection unit 402 selects the QP value of the upper block A. .

  FIG. 10B is a diagram illustrating an example of the QP prediction process (part 2) according to the first embodiment. As shown in FIG. 10B, when the processing block X is other than the head block, the QP selection unit 402 selects the QP value of the block B processed immediately before.

  Here, since the QP value of the block close to the processing block X is used as the QP predicted value, the prediction efficiency of the QP value hardly decreases compared to the case where the immediately preceding QP value is used in the raster order.

  In the above example, the QP prediction value is generated using the QP value of the block processed immediately before except at the head of the block line. However, if the prediction is made from the block adjacent to the processing block, the QP value is generated by another method. May be.

  Further, the processing of the QP selection unit 402 may be the following processing. When the block line to be processed is the 2nd to Nth block lines, the QP value stored in the immediately preceding QP storage unit 401 is a slice QP value, which is a value determined by slice unit. Therefore, it is considered that the prediction efficiency is not so good to adopt the Slice QP value as the QP prediction value of each block.

  Therefore, the QP selection unit 402 selects the QP value acquired by the upper QP acquisition unit 403 when the first block of the 2nd to Nth block lines is processed, and otherwise, the immediately preceding QP You may select the QP value which the memory | storage part 401 hold | maintained.

  Further, the QP selection unit 402 selects the QP value held in the immediately preceding QP storage unit 401 when processing the first block of the (N + 1) th and subsequent block lines. The QP value held by the immediately preceding QP storage unit 401 at this time is the QP value of the last block (also referred to as the last block) of the block line located on the N blocks of the processing block.

  Since the Kth block line and the K + Nth block line are processed by the same decoding processing unit, QP prediction unit, and entropy decoding unit, the block processed immediately before the processing block of the Kth block line is K -The last block of the Nth block line.

  Furthermore, the QP selection unit 402 may always adopt the QP value stored in the immediately preceding QP storage unit 401 in the first block of the block line. In addition, when the block line to be processed is the first block of the 1st to Nth block lines, the QP selection unit 402 may set the QP predicted value as a slice QP value.

  At this time, similarly, the QP prediction value of the first block of the (N + 1) th block line may be the QP value of the last block of the block line located on the N block.

  With the above configuration, the process for obtaining the predicted value of the quantization parameter can be performed in parallel for each block line.

<Operation>
Next, the operation of the video decoding device 10 in the first embodiment will be described. FIG. 11 is a flowchart illustrating an example of the block decoding process according to the first embodiment. The block decoding process shown in FIG. 11 is a one-block process.

  In step S101, each entropy decoding unit entropy decodes encoded information such as a differential motion vector, a QP difference value, and a quantized orthogonal transform coefficient of each block from the input stream. The entropy decoded information is output to a decoding processing unit corresponding to the processed entropy decoding unit.

  In step S102, the motion vector restoration unit 304 of the decoding processing unit acquires the motion vector information of the neighboring blocks from the block information storage unit 244, and calculates a prediction vector.

  In step S103, the motion vector restoration unit 304 restores the motion vector by adding the difference motion vector and the prediction vector. Information on the restored motion vector is stored in the block information storage unit 244.

  In step S104, the predicted pixel generation unit 305 acquires pixel data of the reference picture indicated by the motion vector from the decoded pixel storage unit 105 in which previously decoded pictures are stored, and generates a predicted pixel.

  In step S <b> 105, each QP prediction unit generates a QP prediction value of the processing block and inputs the QP prediction value to the QP restoration unit 301.

  In step S106, the QP restoration unit 301 restores the QP value from the input QP predicted value and QP difference value. The QP value is input to the inverse quantization unit 302 and stored in the block information storage unit 244. By making the QP predicted value the QP value of the upper block of the processing block or the block processed by the block line decoding processing unit 201 in the past, it is not necessary to wait for the generation of the QP predicted value, and the efficiency of parallel processing is improved. To do.

  In step S107, the inverse quantization unit 302 multiplies the quantized orthogonal transform coefficient by the QP value.

  In step S108, the inverse orthogonal transform unit 303 performs an inverse orthogonal transform process on the inversely quantized orthogonal transform coefficient to generate a prediction error pixel.

  In step S109, the decoded pixel generation unit 306 adds the prediction error pixel and the prediction pixel to generate a decoded pixel.

  In step S110, the decoded pixel storage unit 105 stores the decoded pixel. With the above processing, the block decoding process ends, and the process proceeds to the next block decoding process. When the decoding process for all the blocks included in one picture is completed, the decoded image stored in the decoded pixel storage unit 105 is output to a display unit such as a display.

  As described above, according to the first embodiment, in the video decoding device, it is possible to parallelize the calculation processing of the QP prediction value, and to improve the efficiency of the block line parallel processing.

[Example 2]
Next, the moving picture coding apparatus according to the second embodiment will be described. In the second embodiment, corresponding to the first embodiment, QP prediction processing is parallelized in units of block lines.

<Configuration>
FIG. 12 is a block diagram illustrating an example of a configuration of the moving image encoding device 50 according to the second embodiment. In the example illustrated in FIG. 12, the video decoding device 50 includes a parallel encoding unit 501 and a decoded pixel storage unit 505. The parallel encoding unit 501 includes a parallel encoding processing unit 502, a parallel QP prediction unit 503, and a parallel entropy encoding unit 504. Note that the video encoding device 50 performs parallel processing on N block lines.

  Each unit of the parallel encoding unit 501 uses the same block as a block to be processed in each block line. Further, for example, the horizontal position of the processing block of the (K-1) th block line is two blocks or more ahead of the horizontal position of the processing block of the Kth block line.

  This is because the horizontal position is shifted by 2 blocks or more between the block lines, so that even if each block line is encoded in parallel, the encoding information of the upper and upper right blocks can be used. It is to make it. When using decoding information of only the upper block, the horizontal position shift amount between the block lines may be one block.

  The parallel encoding unit 501 divides an image into a plurality of blocks, and performs parallel processing using, for example, N block lines using a moving image encoding method. The decoded pixels locally decoded by the parallel encoding unit 501 are stored in the decoded pixel storage unit 505.

  The parallel encoding processing unit 502 generates a quantized orthogonal transform coefficient, differential motion vector information, and the like in parallel for each block included in the block line. The generated orthogonal transform coefficient and differential motion vector information (also referred to as encoded data) are output to the parallel entropy encoding unit 504. The QP value used in the quantization is output to the parallel QP prediction unit 503.

  The parallel QP prediction unit 503 calculates the prediction value (QP prediction value) of the quantization parameter (QP) of each block included in the block line in parallel for each block line. The calculated QP prediction value is output to parallel entropy encoding unit 504.

  The parallel entropy encoding unit 504 applies entropy to each block included in the block line using a quantized orthogonal transform coefficient, a QP difference value that is a difference between the QP value and the QP prediction value, difference motion vector information, and the like. Encoding is performed in parallel on a block line basis. The stream encoded by the parallel entropy encoding unit 504 is output to the video decoding device 10 or the like.

  The decoded pixel storage unit 505 stores a local decoded pixel of each block output from the parallel encoding unit 501. Local decoding is also called local decoding.

《Parallel coding unit》
Next, the parallel encoding unit 501 will be described. FIG. 13 is a block diagram illustrating an example of a configuration of each unit of the parallel encoding unit 501 according to the second embodiment. In the example illustrated in FIG. 13, the parallel encoding processing unit 502 includes a first encoding processing unit 621, a second encoding processing unit 622, an Nth encoding processing unit 623, and a block information storage unit 624. .

  In the example illustrated in FIG. 13, the parallel QP prediction unit 503 includes a first QP prediction unit 631, a second QP prediction unit 632, and an NQP prediction unit 633. In the example illustrated in FIG. 13, the parallel entropy encoding unit 504 includes a first entropy encoding unit 641, a second entropy encoding unit 642, and an Nth entropy encoding unit 643.

  Here, for L = 1 to N, the L-th encoding processing unit, the LQP prediction unit, and the L-th entropy encoding unit process one block line correspondingly. Hereinafter, the L-th encoding processing unit, the LQP prediction unit, and the L-th entropy encoding unit are collectively referred to as a block line encoding processing unit.

  For example, the block line encoding processing unit 601 includes a first encoding processing unit 621, a first QP prediction unit 631, and a first entropy encoding unit 641.

  When the block line encoding processing unit encodes the Kth block line, the block line encoding processing unit next encodes the K + Nth block line.

  The parallel encoding processing unit 502 divides the input image into a plurality of block lines. Each encoding process part 621-623 performs an encoding process in parallel per block line. The encoding process is, for example, H.264. H.264 or HEVC.

  FIG. 14 is a block diagram illustrating an example of the configuration of the encoding processing unit according to the second embodiment. Since each of the encoding processing units 621 to 623 performs the same processing, the following description will be given using the first encoding processing unit 621.

  The first encoding processing unit 621 includes a prediction difference unit 701, an orthogonal transform unit 702, a quantization unit 703, a QP determination unit 704, an inverse quantization unit 705, an inverse orthogonal transform unit 706, and a decoded pixel generation. Unit 707, motion detection unit 708, prediction signal generation unit 709, and difference vector generation unit 710.

  The motion detection unit 708 acquires pixel data of the reference picture from the decoded pixel storage unit 505, and detects a motion vector. Information on the detected motion vector is stored in the block information storage unit 624 and used for encoding the next block.

  The prediction signal generation unit 709 acquires a reference pixel from the decoded pixel storage unit 505 based on the input region position information of the reference image, and generates a prediction pixel signal. The generated prediction pixel signal is output to the prediction difference unit 701.

  The difference vector generation unit 710 generates a prediction vector. As an example of the prediction vector, the motion vectors of the blocks located on the left, top, and top right of the processing block are acquired from the block information storage unit 624, and an intermediate value among the three motion vectors is used as the prediction vector.

  As described above, each block line is processed with a shift of, for example, two blocks horizontally, so that the encoding processing of the upper and upper right blocks of the processing block is completed. Therefore, the difference vector generation unit 710 can acquire the motion vectors of the peripheral blocks.

  The difference vector generation unit 710 acquires the motion vector of the processing block from the motion detection unit 708, and generates a difference vector between the motion vector and the prediction vector. The generated difference vector is output to the first entropy encoding unit 641.

  The prediction difference unit 701 generates a prediction error signal by calculating a difference between the original image and the prediction pixel signal. The prediction error signal is output to the orthogonal transform unit 702.

  The orthogonal transform unit 702 performs orthogonal transform processing such as discrete cosine transform on the prediction error signal. The transformed orthogonal transform coefficient is output to the quantization unit 703.

  The quantization unit 703 quantizes the orthogonal transform coefficient based on the quantization parameter (QP) value. As an example of the quantization method, the orthogonal transform coefficient is divided by a value determined by QP and rounded to an integer value. If the quantized orthogonal transform coefficient is multiplied by the QP value, inverse quantization can be performed. By the rounding process, quantization becomes an irreversible transformation. The quantized orthogonal transform coefficient is output to the first entropy encoding unit 641.

  Also, the quantization unit 703 generates flag information indicating whether or not the quantized orthogonal transform coefficients are all 0, and outputs the flag information to the QP determination unit 704 together with the QP value used in the quantization. The QP value is output to the inverse quantization unit 705 and the first QP prediction unit 631.

  The first QP prediction unit 631 generates a QP prediction value for the processing block. The generated QP prediction value is output to the QP determination unit 704 and the first entropy encoding unit 641.

  The QP prediction value is set to the upper block of the processing block or the QP value of the block processed in the past by the block line coding processing unit to which the first QP prediction unit 631 belongs, and the generation of the QP prediction value is awaited. This eliminates the need for parallel processing and improves the efficiency of parallel processing.

  The QP determination unit 704 determines the QP of the processing block from the QP value input from the quantization unit 703 and the QP predicted value. When all the orthogonal transform coefficients are 0, the QP difference information is not entropy-coded, and therefore the QP value used in the quantization unit 703 is invalid because it is not notified to the decoding device side.

  For example, the QP determination unit 704 acquires flag information indicating whether or not the quantized orthogonal transform coefficients generated by the quantization unit 703 are all zero. When the QP determination unit 704 acquires flag information indicating all 0s, the QP value of the processing block is set to the QP predicted value. When the QP determination unit 704 acquires flag information indicating that all are not 0, the QP value of the processing block is set to the QP value used in the quantization unit 703. The QP value determined by the QP determination unit 704 is stored in the block information storage unit 624.

  The inverse quantization unit 705 performs an inverse quantization process on the quantized orthogonal transform coefficient. The inversely quantized orthogonal transform coefficient is output to the inverse orthogonal transform unit 706.

  The inverse orthogonal transform unit 706 performs an inverse orthogonal transform process on the inversely quantized orthogonal transform coefficient. The inverse orthogonal transformed signal is output to the decoded pixel generation unit 707.

  The decoded pixel generation unit 707 generates a local decoded pixel by adding the prediction pixel signal acquired from the prediction signal generation unit 709 to the signal subjected to inverse orthogonal transform. The generated decoded pixel is stored in the decoded pixel storage unit 505.

  Next, processing of each QP prediction unit will be described. FIG. 15 is a block diagram illustrating an example of a configuration of a QP prediction unit according to the second embodiment. Since each of the QP prediction units 631 to 633 performs the same processing, the following description will be given using the first QP prediction unit 631.

  The first QP prediction unit 631 illustrated in FIG. 15 includes a previous QP storage unit 801, a QP selection unit 802, and an upper QP acquisition unit 803.

  The immediately preceding QP storage unit 801 receives the QP value of the block processed immediately before the current processing block from the QP determination unit 704 and stores it. The QP value stored in the immediately preceding QP storage unit 801 is initialized at the start of picture processing.

  For example, H.M. Similarly to H.264, initialization is performed with the Slice QP value encoded in the Slice header information. Slice is a unit when a block belonging to one picture is divided into groups.

  The upper QP acquisition unit 803 acquires the QP value of the block located above the processing block from the block information storage unit 624.

  The QP selection unit 802 selects the QP value output from either the immediately preceding QP storage unit 801 or the upper QP acquisition unit 803, and uses the QP determination unit 704 or the first entropy encoding unit 641 as a QP predicted value. Output to.

  For example, the QP selection unit 802 selects the QP value output from the upper QP acquisition unit 803 if the processing block is the block block head block, and if the processing block is a block other than the block line head block, The QP value output from the storage unit 801 is selected. The QP selection unit 802 outputs the selected QP value as a QP predicted value to the QP determination unit 704 and the first entropy encoding unit 641.

  The prediction (selection) of the QP value in the above example is as shown in FIGS. 10A and 10B. Here, since the QP value of the block close to the processing block X is used as the QP predicted value, the prediction efficiency of the QP value hardly decreases compared to the case where the immediately preceding QP value is used in the raster order.

  In the above example, the QP prediction value is generated using the QP value of the block processed immediately before except at the head of the block line. However, if the prediction is made from the block adjacent to the processing block, the QP value is generated by another method. May be.

  As in the first embodiment of the video decoding device 10, the processing of the QP selection unit 802 may be as follows. When the block line to be processed is the 2nd to Nth block lines, the QP value stored in the immediately preceding QP storage unit 801 is a slice QP value, which is a value determined by slice unit. Therefore, when the slice QP value is adopted as the QP prediction value of each block, the prediction efficiency may not be so good.

  Therefore, the QP selection unit 802 selects the QP value acquired by the upper QP acquisition unit 803 when the block line to be processed processes the first block of the 2nd to Nth block lines, and otherwise stores the immediately preceding QP storage. The QP value held by the unit 801 may be selected.

  Further, the QP selection unit 802 selects the QP value held by the immediately preceding QP storage unit 801 when processing the first block of the N + 1th and subsequent block lines. The QP value held by the immediately preceding QP storage unit 801 at this time is the QP value of the last block on the block line located on the N block of the processing block.

  Since the Kth block line and the K + Nth block line are processed by the same decoding processing unit, QP prediction unit, and entropy decoding unit, the block processed immediately before the processing block of the Kth block line is K -The last block of the Nth block line.

  Furthermore, the QP selection unit 802 may always adopt the QP value stored in the immediately preceding QP storage unit 801 in the first block of the block line. Further, when the block line to be processed is the first block of the 1st to Nth block lines, the QP selection unit 802 may set the QP predicted value as a slice QP value.

  At this time, similarly, the QP predicted value of the first block of the (N + 1) th block line may be the QP value of the last block (final block) of the block line located on the N block.

  Each entropy encoding unit 641 to 643 entropy encodes the differential motion vector, QP difference value, quantized orthogonal transform coefficient, and the like of each block for each block line.

  With the above configuration, the process for obtaining the predicted value of the quantization parameter can be performed in parallel for each block line.

<Operation>
Next, the operation of the moving picture coding apparatus 50 according to the second embodiment will be described. FIG. 16 is a flowchart illustrating an example of a block encoding process according to the second embodiment. The block encoding process shown in FIG. 16 is a one-block process.

  In step S201, a block to be processed (processing block) is input to each encoding processing unit. The motion detection unit 708 acquires pixel data of the reference picture from the decoded pixel storage unit 505, and detects a motion vector.

  In step S202, the difference vector generation unit 710 generates a difference vector between the detected motion vector and the generated prediction vector.

  In step S203, the motion detection unit 708 stores the detected motion vector in the block information storage unit 624.

  In step S204, the prediction signal generation unit 709 acquires region position information of a reference image that is referenced by the motion vector detected by the motion detection unit 708, and generates a prediction pixel signal.

  In step S205, the prediction difference unit 701 takes the difference between the prediction pixel signal generated by the prediction signal generation unit 709 and the input original image, and generates a prediction error signal.

  In step S206, the orthogonal transform unit 702 performs orthogonal transform processing on the prediction error signal generated by the prediction difference unit 701 to generate orthogonal transform coefficients.

  In step S207, the quantization unit 703 quantizes the orthogonal transform coefficient based on the quantization parameter (QP) value.

In step S208, each QP prediction unit generates a QP prediction value for the processing block. The QP prediction value is, for example, the QP value of the upper block of the processing block or the block processed in the past by the block ry encoding unit to which the QP prediction unit belongs.

  In step S209, the QP determination unit 704 determines the QP value of the processing block as one of the QP value acquired from the quantization unit 703 and the QP prediction value acquired from the QP prediction unit. The method of determination is as described above.

  In step S210, each entropy coding unit entropy codes the quantized orthogonal transform coefficient, QP difference value, difference vector, and the like of the processing block.

  In step S211, the inverse quantization unit 705 and the inverse orthogonal transform unit 706 perform an inverse quantization process on the quantized orthogonal transform coefficient, perform an inverse orthogonal transform process, and a signal corresponding to the prediction error signal. Is generated.

  In step S212, the decoded pixel generation unit 707 adds the prediction pixel signal acquired from the prediction signal generation unit 709 and the signal acquired from the inverse orthogonal transform unit 706, and generates a local decoded pixel.

  In step S213, the decoded pixel storage unit 505 stores the generated decoded pixel. With the above process, the block encoding process is completed, and the process proceeds to the next block encoding process.

  As described above, according to the second embodiment, it is possible to parallelize the calculation processing of the QP prediction value in the video encoding device, and to improve the efficiency of the block line parallel processing.

[Example 3]
FIG. 17 is a block diagram illustrating an example of the configuration of the moving image processing apparatus. The moving image processing device 90 is an example of the moving image encoding device or the moving image decoding device described in each embodiment. As illustrated in FIG. 17, the moving image processing apparatus 90 includes a control unit 901, a main storage unit 902, an auxiliary storage unit 903, a drive device 904, a network I / F unit 906, an input unit 907, and a display unit 908. These components are connected to each other via a bus so as to be able to transmit and receive data.

  The control unit 901 is a CPU that controls each device, calculates data, and processes in a computer. The control unit 901 is an arithmetic device that executes programs stored in the main storage unit 902 and the auxiliary storage unit 903. The control unit 901 receives data from the input unit 907 and the storage device, calculates, and processes the data, and then displays the display unit 908. Or output to a storage device.

  The main storage unit 902 is a ROM (Read Only Memory), a RAM (Random Access Memory), or the like, and a storage device that stores or temporarily stores programs and data such as an OS and application software that are basic software executed by the control unit 901. It is.

  The auxiliary storage unit 903 is an HDD (Hard Disk Drive) or the like, and is a storage device that stores data related to application software or the like.

  The drive device 904 reads the program from the recording medium 905, for example, a flexible disk, and installs it in the storage device.

  The recording medium 905 stores a predetermined program. The program stored in the recording medium 905 is installed in the moving image processing apparatus 90 via the drive apparatus 904. The installed predetermined program can be executed by the moving image processing apparatus 90.

  The network I / F unit 906 is a peripheral having a communication function connected via a network such as a LAN (Local Area Network) or a WAN (Wide Area Network) constructed by a data transmission path such as a wired and / or wireless line. This is an interface between the device and the moving image processing apparatus 90.

  The input unit 907 includes a keyboard having cursor keys, numeric input, various function keys, and the like, and a mouse and a slice pad for performing key selection on the display screen of the display unit 908. The input unit 907 is a user interface for a user to give an operation instruction to the control unit 901 or input data.

  The display unit 908 includes an LCD (Liquid Crystal Display) or the like, and performs display according to display data input from the control unit 901. Note that the display unit 908 may be provided outside, and in that case, the moving image processing apparatus 90 includes a display control unit.

  As described above, the moving image encoding process or the moving image decoding process described in the above-described embodiment may be realized as a program for causing a computer to execute. By installing this program from a server or the like and causing the computer to execute it, the above-described moving image encoding process or moving image decoding process can be realized.

  Further, the moving image encoding program or the moving image decoding program is recorded on the recording medium 905, and the recording medium 905 on which the program is recorded is read by a computer or a portable terminal, so that the above-described moving image encoding process or moving image is performed. Decoding processing can also be realized.

  The recording medium 905 is a recording medium that records information optically, electrically, or magnetically, such as a CD-ROM, flexible disk, magneto-optical disk, etc., and information electrically, such as a ROM or flash memory. Various types of recording media such as a semiconductor memory for recording can be used. Note that the recording medium 805 does not include a carrier wave.

  The program executed by the moving image processing apparatus 90 has a module configuration including each unit described in each embodiment. As actual hardware, when the control unit 901 reads and executes a program from the auxiliary storage unit 903, one or more of the above-described units are loaded on the main storage unit 902, and one or more of the units are main. It is generated on the storage unit 902.

  Further, the moving picture encoding process or the moving picture decoding process described in each of the above embodiments may be implemented in one or a plurality of integrated circuits.

  Although each embodiment has been described in detail above, it is not limited to a specific embodiment, and various modifications and changes can be made within the scope described in the claims. It is also possible to combine all or a plurality of the components of the above-described embodiments.

In addition, the following additional remarks are disclosed regarding the above Example.
(Appendix 1)
A video decoding device for decoding a stream encoded using a video encoding method for an image divided into a plurality of blocks,
A parallel entropy decoding unit that entropy-decodes a stream of blocks included in a block line indicating the row of blocks in parallel for each block line;
A parallel QP prediction unit that calculates a prediction value of a quantization parameter of each block included in the block line in parallel for each block line;
For each block included in the block line, a decoded pixel decoded using the data decoded by the parallel entropy decoding unit and the prediction value calculated by the parallel QP prediction unit is parallelized for each block line. A parallel decoding processing unit to generate,
The parallel QP prediction unit includes:
When parallel processing is performed in units of N block lines each having a value of 2 or more, the horizontal position of the processing block of the (K-1) th block line is at least one block ahead of the horizontal position of the processing block of the Kth block line. A video decoding apparatus that calculates the prediction value with reference to a block that has been processed in the parallel processing with respect to the processing block.
(Appendix 2)
The parallel QP prediction unit includes:
The moving picture decoding apparatus according to supplementary note 1, wherein when the first block of the second and subsequent block lines from the top of the image is processed, a block one block above the first block is referred to.
(Appendix 3)
The parallel QP prediction unit includes:
When processing the first block of the 2nd to Nth block lines from the top of the image, refer to the block one block above the first block, and when processing the first block of the N + 1th and subsequent block lines, the start block The moving picture decoding apparatus according to supplementary note 1, which refers to a final block in a block line on N blocks of the block.
(Appendix 4)
The predicted value is
When processing the first block of the 1st to Nth block lines from the top of the image, the quantization parameter value included in the header information of the image, and when processing the first block of the N + 1 and subsequent block lines, The moving picture decoding apparatus according to appendix 1, wherein the quantization parameter value of the last block in the block line on the N blocks of the first block is used.
(Appendix 5)
A moving image encoding apparatus that divides an image into a plurality of blocks and performs a moving image encoding process for each block,
A parallel encoding processing unit that generates, in parallel for each block line, encoded data including quantized orthogonal transform coefficients for blocks included in the block line indicating the row of blocks;
A parallel QP prediction unit that calculates a prediction value of a quantization parameter of each block included in the block line in parallel for each block line;
A parallel entropy encoding unit that performs entropy encoding using the encoded data and the prediction value for each block included in the block line in parallel for each block line, and
The parallel QP prediction unit includes:
When parallel processing is performed in units of N block lines each having a value of 2 or more, the horizontal position of the processing block of the (K-1) th block line is at least one block ahead of the horizontal position of the processing block of the Kth block line. A video encoding apparatus that calculates the prediction value with reference to a block that has been processed in the parallel processing with respect to the processing block.
(Appendix 6)
A moving image decoding method executed by a moving image decoding apparatus for decoding a stream encoded using a moving image encoding method for an image divided into a plurality of blocks,
A block stream included in a block line indicating a row of the blocks is entropy-decoded in parallel for each block line,
The prediction value of the quantization parameter of each block included in the block line is calculated in parallel for each block line,
For each block included in the block line, a process of generating decoded pixels decoded using the data decoded by the parallel entropy decoding process and the predicted value in parallel for each block line,
The process of calculating the predicted value includes
When parallel processing is performed in units of N block lines each having a value of 2 or more, the horizontal position of the processing block of the (K-1) th block line is at least one block ahead of the horizontal position of the processing block of the Kth block line. The video decoding method of calculating the prediction value with reference to a block that has been processed in the parallel processing with respect to the processing block.
(Appendix 7)
A moving image encoding method executed by a moving image encoding apparatus that divides an image into a plurality of blocks and performs a moving image encoding process for each block,
For block included in the block line indicating the row of the block, the encoded data including the quantized orthogonal transform coefficient is generated in parallel for each block line,
The prediction value of the quantization parameter of each block included in the block line is calculated in parallel for each block line,
For each block included in the block line, a process of performing entropy encoding using the encoded data and the predicted value in parallel for each block line,
The process of calculating the predicted value includes
When parallel processing is performed in units of N block lines each having a value of 2 or more, the horizontal position of the processing block of the (K-1) th block line is at least one block ahead of the horizontal position of the processing block of the Kth block line. The moving picture coding method of calculating the prediction value with reference to a block that has been processed in the parallel processing with respect to the processing block.
(Appendix 8)
A moving picture decoding program for causing a moving picture decoding apparatus that decodes a stream encoded using a moving picture encoding scheme to an image divided into a plurality of blocks,
A block stream included in a block line indicating a row of the blocks is entropy-decoded in parallel for each block line,
The prediction value of the quantization parameter of each block included in the block line is calculated in parallel for each block line,
For each block included in the block line, a process of generating decoded pixels decoded using the data decoded by the parallel entropy decoding process and the predicted value in parallel for each block line,
The process of calculating the predicted value includes
When parallel processing is performed in units of N block lines each having a value of 2 or more, the horizontal position of the processing block of the (K-1) th block line is at least one block ahead of the horizontal position of the processing block of the Kth block line. A moving picture decoding program that calculates the prediction value with reference to a block that has been processed in the parallel processing with respect to the processing block.
(Appendix 9)
A moving picture coding program for dividing a picture into a plurality of blocks and causing a moving picture coding apparatus to perform a moving picture coding process for each block.
For block included in the block line indicating the row of the block, the encoded data including the quantized orthogonal transform coefficient is generated in parallel for each block line,
The prediction value of the quantization parameter of each block included in the block line is calculated in parallel for each block line,
For each block included in the block line, a process of performing entropy encoding using the encoded data and the predicted value in parallel for each block line,
The process of calculating the predicted value includes
When parallel processing is performed in units of N block lines each having a value of 2 or more, the horizontal position of the processing block of the (K-1) th block line is at least one block ahead of the horizontal position of the processing block of the Kth block line. A moving picture coding program that calculates the prediction value with reference to a block that has been processed by the parallel processing with respect to the processing block.

DESCRIPTION OF SYMBOLS 10 Moving picture decoding apparatus 50 Moving picture encoding apparatus 90 Moving picture processing apparatus 101 Parallel decoding part 102 Parallel entropy decoding part 103 Parallel QP prediction part 104 Parallel decoding processing part 105 Decoded pixel memory | storage part 201 Block line decoding process part 244 Block information storage Unit 301 QP restoration unit 401 immediately preceding QP storage unit 402 QP selection unit 403 upper QP acquisition unit 501 parallel coding unit 502 parallel coding processing unit 503 parallel QP prediction unit 504 parallel entropy coding unit 505 decoding pixel storage unit 601 block line code Processing unit 624 Block information storage unit 704 QP determination unit 801 Immediately before QP storage unit 802 QP selection unit 803 Upper QP acquisition unit 901 Control unit 902 Main storage unit 903 Auxiliary storage unit

Claims (2)

A moving image processing apparatus including a moving image encoding unit that divides an image into a plurality of blocks and performs a moving image encoding process for each block, and a moving image decoding unit that decodes an encoded stream. ,
The moving image encoding unit includes:
A parallel encoding processing unit that generates, in parallel for each block line, encoded data including quantized orthogonal transform coefficients for blocks included in the block line indicating the row of blocks;
A first parallel QP prediction unit that calculates a prediction value of a quantization parameter for a quantization parameter used in encoding of each block included in the block line in parallel for each block line;
A parallel entropy encoding unit that performs entropy encoding on each block included in the block line using the encoded data and the predicted value of the quantization parameter in parallel for each block line, and
The first parallel QP prediction unit includes:
When parallel processing is performed in units of N block lines each having a value of 2 or more, the horizontal position of the processing block of the (K-1) th block line is at least one block ahead of the horizontal position of the processing block of the Kth block line. To process
When the processing block is at the head of the block line, the quantization parameter of the block processed immediately before the processing block is initialized with the quantization parameter determined in units of the slice to which the processing block belongs, and the initialized quantization parameter is set. A predicted value of the quantization parameter of the processing block based on
In the case where the processing block is other than the head of the block line, the predicted value of the quantization parameter of the processing block is calculated based on the quantization parameter of the block in the block line,
The parallel encoding processing unit
The quantization parameter of the processing block based on the prediction value of the quantization parameter calculated by the first parallel QP prediction unit, the quantization parameter used for the processing of each block, and whether the result of encoding is all 0 or not Decide
The moving image decoding unit
A parallel entropy decoding unit that entropy-decodes a stream of blocks included in a block line indicating the row of blocks in parallel for each block line;
A second parallel QP prediction unit that calculates a prediction value of a quantization parameter of each block included in the block line in parallel for each block line;
For each block included in the block line, a decoded pixel obtained by performing decoding processing using the data decoded by the parallel entropy decoding unit and the prediction value calculated by the second parallel QP prediction unit, A parallel decoding processing unit that generates each in parallel,
The second parallel QP prediction unit includes:
When parallel processing is performed for each of N block lines indicating a value of 2 or more, the relationship between the horizontal position of the processing block of the (K-1) th block line and the horizontal position of the processing block of the Kth block line is K-1 The timing at which the horizontal position of the processing block of the first block line precedes at least one block,
When the processing block is at the head of the block line, the quantization parameter of the block processed immediately before the processing block is initialized with the quantization parameter determined in units of the slice to which the processing block belongs, and the initialized quantization parameter is set. A predicted value of the quantization parameter of the processing block based on
In the case where the processing block is other than the head of the block line, the prediction value of the quantization parameter of the processing block is calculated based on the quantization parameter of the block that has been decoded earlier in the block line,
The parallel decoding processing unit
A moving image that calculates a quantization parameter of a processing block based on the predicted value of the quantization parameter calculated by the second parallel QP prediction unit, and performs a deblocking filter process of the processing block using the calculated quantization parameter Processing equipment. A moving image of a moving image processing apparatus having a moving image encoding unit that divides an image into a plurality of blocks and performs a moving image encoding process for each block, and a moving image decoding unit that decodes the encoded stream A processing method,
The moving image encoding unit includes:
For block included in the block line indicating the row of the block, the encoded data including the quantized orthogonal transform coefficient is generated in parallel for each block line,
The prediction value of the quantization parameter for the quantization parameter used in the encoding of each block included in the block line is calculated in parallel for each block line,
For each block included in the block line, performing entropy coding using the encoded data and the predicted value of the quantization parameter in parallel for each block line,
The process of calculating the predicted value includes
When parallel processing is performed in units of N block lines each having a value of 2 or more, the horizontal position of the processing block of the (K-1) th block line is at least one block ahead of the horizontal position of the processing block of the Kth block line. To process
When the processing block is at the head of the block line, the quantization parameter of the block processed immediately before the processing block is initialized with the quantization parameter determined in units of the slice to which the processing block belongs, and the initialized quantization parameter is set. A predicted value of the quantization parameter of the processing block based on
In the case where the processing block is other than the head of the block line, the predicted value of the quantization parameter of the processing block is calculated based on the quantization parameter of the block in the block line,
The process of generating the encoded data includes
Based on the prediction value of the quantization parameter calculated in the process of calculating the prediction value, the quantization parameter used for the processing of each block, and whether the result of the encoding is all 0 or not, the quantization parameter of the processing block is determined. Decide
The moving image decoding unit
A block stream included in a block line indicating a row of the blocks is entropy-decoded in parallel for each block line,
The prediction value of the quantization parameter of each block included in the block line is calculated in parallel for each block line,
For each block included in the block line, a process of generating decoded pixels decoded using the parallel entropy-decoded data and the predicted value for each block line is executed.
The process of calculating the predicted value includes
When parallel processing is performed for each of N block lines indicating a value of 2 or more, the relationship between the horizontal position of the processing block of the (K-1) th block line and the horizontal position of the processing block of the Kth block line is K-1 The timing at which the horizontal position of the processing block of the first block line precedes at least one block,
When the processing block is at the head of the block line, the quantization parameter of the block processed immediately before the processing block is initialized with the quantization parameter determined in units of the slice to which the processing block belongs, and the initialized quantization parameter is set. A predicted value of the quantization parameter of the processing block based on
In the case where the processing block is other than the head of the block line, the prediction value of the quantization parameter of the processing block is calculated based on the quantization parameter of the block that has been decoded earlier in the block line,
The process of generating the decoded pixel includes:
Video processing for calculating a quantization parameter of a processing block based on a prediction value of a quantization parameter calculated in the process of calculating the prediction value, and performing deblocking filter processing of the processing block using the calculated quantization parameter Method.

Images