JP2013141296A - Image decoding apparatus, image decoding method, image decoding program, reception device, reception method, and reception program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve encoding efficiency of motion information while suppressing a load when processing the motion information.SOLUTION: When an inter-image prediction mode indicated by prediction mode information is a first inter-image prediction mode using motion vector information included in encoding information of a reference block specified by an index, a first mode output section specifies the reference block of a decoding target block on the basis of priority and the index of the reference block and outputs the encoding information. When the inter-image prediction mode indicated by the prediction mode information is a second inter-image prediction mode using a differential motion vector corresponding to the reference block specified by the index, a second mode output section specifies the reference block of the decoding target block on the basis of the priority and the index of the reference block, and adds a prediction motion vector based on the motion vector information included in the encoding information of the reference block and the differential motion vector to calculate motion vector of the decoding target block and outputs the calculated motion vector of the decoding target block.

Description

  The present invention relates to a moving picture coding technique, and more particularly to an image decoding apparatus, an image decoding method, an image decoding program, a receiving apparatus, and a receiving method that divide a picture into rectangular blocks and perform motion estimation and compensation between blocks. And a receiving program.

  In a moving picture coding method represented by MPEG (Moving Picture Coding Experts Group), which divides a picture into rectangular blocks and performs motion estimation and compensation between blocks, the coding of motion vectors generated in each block In order to reduce the amount, a prediction process is performed on the motion vector.

  In MPEG-2, the motion vector detected in units of macroblocks is taken as a difference from the motion vector of the macroblock encoded immediately before, and the amount of code is reduced by encoding the difference vector. Yes. MPEG-4 AVC / H. H.264 uses the fact that motion vectors have a strong correlation with the motion vectors of neighboring blocks, and performs prediction from neighboring blocks and encodes the difference vector to reduce the amount of code. . Specifically, the median is calculated from the motion vectors of adjacent blocks on the left, top and top right of the block to be processed, and motion vector prediction is realized by taking the difference from the median.

  In these prediction methods, since there is only one motion vector for prediction, there is a problem that if the prediction is not successful, the difference between the motion vectors becomes large and the generated code amount increases. Further, although the amount of code of the motion vector is reduced, other motion information is encoded for each block to be processed. Therefore, even if it has the same motion information as neighboring neighboring blocks, it is encoded redundantly, so that there is a problem that efficient encoding has not been achieved.

  In order to solve these problems, two new technologies have been studied in the standard work of moving picture coding in ISO / IEC and ITU-T. One is a motion vector prediction method, in which a motion vector of an adjacent neighboring block that has been encoded and a motion vector of a peripheral block at the same position of another picture that has different encoding times are each predicted motion vector candidates This is an evaluation method based on the amount of generated codes when applied as.

  FIGS. 1A and 1B show an example of adjacent blocks that are candidates for a motion vector predictor. FIG. 1A is an example of an adjacent block in the same picture, and FIG. 1B is an example of a peripheral block at the same position of another picture having a different time. The motion vectors of these blocks are used as predicted motion vector candidates, and the predicted motion vector that minimizes the generated code amount of the difference value between the motion vector of the block to be processed and the predicted motion vector candidate is selected. And the additional value regarding the adjacent block which selected the difference value with the said prediction motion vector and the prediction motion vector as needed is encoded and transmitted.

  The other is that if the motion information of the processing target block is the same as the motion information of neighboring neighboring blocks that have already been encoded, the processing target block does not encode its own motion information, Information is used for encoding. Specifically, this is a technique of reducing the amount of code of motion information by encoding additional information that designates an adjacent block having motion information to be referenced (see, for example, Patent Document 1). Such a method is called “merge” and has attracted attention as a method for reducing the code amount of motion information.

JP-A-10-276439

  In the motion vector prediction method and the merge method described above, the motion vector of the encoded neighboring neighboring block and the neighboring block at the same position of another picture with different time, the reference picture number indicating the reference picture, the reference list, etc. Use encoded information. However, since the positions of adjacent neighboring blocks to be referred to by the respective technologies are different, there is a difficult aspect that the number of times of accessing the memory storing the encoded information that has been encoded increases.

  In a general encoding process, in order to select an optimal reference destination, motion compensation is performed using the encoding information of the adjacent block of the reference destination, and the generated code amount and the encoding distortion are determined as indices. . However, if adjacent neighboring blocks to be referred to are different, the number of blocks increases, which causes a difficult process load. Furthermore, there is a difficult aspect that the timing for reading out the encoded information at the time of decoding is limited, and the temporary memory for storing the encoded information that has been decoded increases.

  The present invention has been made in view of such a situation, and an object of the present invention is to provide a technique for improving the encoding efficiency of motion information while suppressing the load when processing the motion information.

In order to solve the above problems, an image decoding apparatus according to an aspect of the present invention is an image decoding apparatus that decodes a code string that is encoded using a motion vector in units of blocks obtained by dividing each picture of a moving image. For each decoding target block from the code string, prediction mode information indicating an inter-picture prediction mode and an index of a reference block, or prediction mode information indicating an inter-picture prediction mode, an index of a reference block, and a differential motion vector corresponding to the reference block And a coding information storage unit (210) for storing the coding information of each block, and coding information of a plurality of reference block candidates of the decoding target block. An acquisition unit (205) acquired from the encoded information storage unit (210) and the prediction mode information decoded by the code string decoding unit (202). Is the first inter-picture prediction mode using motion vector information included in the coding information of the reference block specified by the index, the plurality of the prediction modes based on the first predetermined order. A first candidate list is generated from the reference block candidates, and the reference block of the decoding target block is determined from the first candidate list based on the index of the reference block decoded by the code string decoding unit (202). A first mode output unit (206) that specifies and outputs the encoded information, and an inter-picture prediction mode indicated by the prediction mode information decoded by the code string decoding unit (202) is specified by an index In the case of the second inter-picture prediction mode using the difference motion vector corresponding to the block, the plurality of the plurality of the plurality of motion vectors based on the second predetermined order Generating a second candidate list from the reference block candidates, and identifying the reference block of the decoding target block from the second candidate list based on the index of the reference block decoded by the code string decoding unit, Calculating and outputting a motion vector of the decoding target block from a motion vector predictor based on motion vector information included in the encoding information of a reference block and a differential motion vector decoded by the code string decoding unit (202); And the code string decoding unit (202) based on the information output from the first mode output unit (206) or the second mode output unit (204). A motion compensation prediction unit (207) that generates a predicted image by performing motion compensation using the inter-picture prediction mode indicated by the predicted prediction mode information. The first mode output unit (206) and the second mode output unit (204) use a plurality of common blocks as the plurality of reference block candidates, and for the plurality of reference block candidates, An already decoded block spatially adjacent to the decoding target block, and a block included in a picture temporally different from the picture including the decoding target block, the first mode output unit (206) and the second The mode output unit (204) of the plurality of reference block candidates has spatially adjacent ranks in the first predetermined rank and the second predetermined rank of blocks included in the temporally different pictures. The block is set lower than the ranks in the first predetermined rank and the second predetermined rank.
Another aspect of the present invention is an image decoding method. This method is an image decoding method for decoding a code string encoded using a motion vector in units of blocks obtained by dividing each picture of a moving image, and includes an inter-picture prediction mode for each decoding target block from the code string. A first step of decoding prediction mode information indicating a reference block index, or prediction mode information indicating an inter-picture prediction mode, a reference block index, and a differential motion vector corresponding to the reference block; and a plurality of the decoding target blocks The second step of obtaining the encoding information of the reference block candidate of the reference, and the inter-picture prediction mode indicated by the prediction mode information decoded by the first step are provided in the encoding information of the reference block specified by the index In the first inter-image prediction mode using vector information, the first predetermined order is used. Generating a first candidate list from the plurality of reference block candidates, and identifying the reference block of the decoding target block from the first candidate list based on the index of the reference block decoded in the first step Then, a third step for outputting the encoded information, and a second step in which the inter-picture prediction mode indicated by the prediction mode information decoded in the first step uses a differential motion vector corresponding to the reference block specified by the index. In the inter-picture prediction mode, a second candidate list is generated from the plurality of reference block candidates based on a second predetermined order, and based on the index of the reference block decoded in the first step Identifying a reference block of the decoding target block from the second candidate list, and encoding the reference block A fourth step of calculating and outputting a motion vector of the block to be decoded from a prediction motion vector based on motion vector information included in the report and a difference motion vector decoded in the first step; and the third step or the first step And a fifth step of generating a predicted image by performing motion compensation using the inter-picture prediction mode indicated by the prediction mode information decoded in the first step based on the information output in the four steps. The third step and the fourth step use a plurality of common blocks as the plurality of reference block candidates, and the plurality of reference block candidates are already decoded spatially adjacent to the decoding target block. And a block included in a picture temporally different from a picture including the decoding target block, wherein the third step and the fourth step are performed to change the temporally different pictures in the plurality of reference block candidates. The ranks of the included blocks in the first predetermined rank and the second predetermined rank are set lower than the ranks in the first predetermined rank and the second predetermined rank of the spatially adjacent blocks.
Another aspect of the present invention is a receiving device. This apparatus is a receiving apparatus that receives and decodes a code sequence in which a moving image is encoded, and the code sequence encoded using a motion vector in units of blocks obtained by dividing each picture of the moving image is packetized. A receiving unit that receives the encoded stream, a restoration unit that restores the original code sequence by packet processing the received packetized encoded stream, and a decoding target from the restored original code sequence Code sequence decoding for decoding, for each block, prediction mode information indicating an inter-picture prediction mode and an index of a reference block, or prediction mode information indicating an inter-picture prediction mode, an index of a reference block, and a differential motion vector corresponding to the reference block Unit (202), an acquisition unit (205) for acquiring encoding information of a plurality of reference block candidates of the decoding target block, and the code string recovery When the inter-picture prediction mode indicated by the prediction mode information decoded by the unit (202) is the first inter-picture prediction mode that uses the motion vector information included in the encoded information of the reference block specified by the index. , Generating a first candidate list from the plurality of reference block candidates based on a first predetermined order, and based on the index of the reference block decoded by the code string decoding unit (202), the first list A first mode output unit (206) that identifies a reference block of the decoding target block from the candidate list and outputs the encoded information, and an image indicated by the prediction mode information decoded by the code string decoding unit (202) The inter prediction mode is the second inter image prediction mode that uses the differential motion vector corresponding to the reference block specified by the index. In addition, a second candidate list is generated from the plurality of reference block candidates based on a second predetermined order, and the second candidate list is generated based on the index of the reference block decoded by the code string decoding unit. A reference block of the decoding target block is identified from the list, and the prediction motion vector based on the motion vector information included in the encoding information of the reference block and the difference motion vector decoded by the code string decoding unit (202) are used. The second mode output unit (204) that calculates and outputs the motion vector of the decoding target block, and the information output from the first mode output unit (206) or the second mode output unit (204) Based on the prediction mode information decoded by the code string decoding unit (202), motion compensation is performed to generate a predicted image. A motion compensation prediction unit (207). The first mode output unit (206) and the second mode output unit (204) use a plurality of common blocks as the plurality of reference block candidates, and for the plurality of reference block candidates, An already decoded block spatially adjacent to the decoding target block, and a block included in a picture temporally different from the picture including the decoding target block, the first mode output unit (206) and the second The mode output unit (204) of the plurality of reference block candidates has spatially adjacent ranks in the first predetermined rank and the second predetermined rank of blocks included in the temporally different pictures. The block is set lower than the ranks in the first predetermined rank and the second predetermined rank.
Another aspect of the present invention is a reception method. This method is a receiving method for receiving and decoding a code sequence in which a moving image is encoded, and the code sequence encoded using a motion vector in units of blocks obtained by dividing each picture of the moving image is packetized. A receiving step for receiving the encoded stream, a restoration step for restoring the original code string by packet processing the received packetized encoded stream, and a decoding target from the restored original code string First step for decoding prediction mode information indicating an inter-picture prediction mode and an index of a reference block, or prediction mode information indicating an inter-picture prediction mode, an index of a reference block, and a differential motion vector corresponding to the reference block for each block A second step of obtaining encoding information of a plurality of reference block candidates of the decoding target block, and the first step. When the inter-picture prediction mode indicated by the prediction mode information decoded by the first prediction mode is the first inter-picture prediction mode using the motion vector information included in the encoded information of the reference block specified by the index, the first inter-picture prediction mode is used. A first candidate list is generated from the plurality of reference block candidates based on a predetermined order, and the decoding target block is decoded from the first candidate list based on the index of the reference block decoded in the first step. A third step of identifying a reference block and outputting its encoded information, and a differential motion vector corresponding to the reference block identified by the index, in which the inter-picture prediction mode indicated by the prediction mode information decoded in the first step In the second inter-picture prediction mode using the plurality of reference blocks based on the second predetermined order. A second candidate list is generated from the first candidate list, the reference block of the decoding target block is identified from the second candidate list based on the index of the reference block decoded in the first step, and the reference A fourth step of calculating and outputting a motion vector of the block to be decoded from a motion vector predictor based on motion vector information included in the encoded information of the block and a differential motion vector decoded in the first step; A fifth step of generating a predicted image by performing motion compensation using the inter-picture prediction mode indicated by the prediction mode information decoded in the first step based on the information output in the third step or the fourth step; . The third step and the fourth step use a plurality of common blocks as the plurality of reference block candidates, and the plurality of reference block candidates are already decoded spatially adjacent to the decoding target block. And a block included in a picture temporally different from a picture including the decoding target block, wherein the third step and the fourth step are performed to change the temporally different pictures in the plurality of reference block candidates. The ranks of the included blocks in the first predetermined rank and the second predetermined rank are set lower than the ranks in the first predetermined rank and the second predetermined rank of the spatially adjacent blocks.
It should be noted that any combination of the above-described constituent elements and a conversion of the expression of the present invention between a method, an apparatus, a system, a recording medium, a computer program, and the like are also effective as an aspect of the present invention.

  ADVANTAGE OF THE INVENTION According to this invention, the encoding efficiency of motion information can be improved, suppressing the load at the time of processing motion information.

FIGS. 1A and 1B are diagrams for explaining an example of a prediction block adjacent to a processing target prediction block and the same picture or another picture having a different time. It is a block diagram which shows the structure of the moving image encoder in the Example of this invention. FIGS. 3A and 3B are diagrams for explaining the definition of the prediction mode classification in the embodiment of the present invention. It is a block diagram which shows the structure of the moving image decoding apparatus in the Example of this invention. It is a figure for demonstrating the definition of the encoding block in the Example of this invention. It is a figure explaining the syntax pattern of the bit stream in the prediction block level regarding the selection method of the reference adjacent block in the Example of this invention. FIGS. 7A to 7D are diagrams for explaining types of the shape of the prediction block in the embodiment of the present invention. It is a figure explaining the preservation | save format of the encoding information recorded on the encoding information storage memory in the Example of this invention. FIGS. 9A and 9B are diagrams illustrating the arrangement of adjacent blocks serving as reference destinations for the prediction block to be processed in the embodiment of the present invention. It is a block diagram which shows the detailed structure of the inter prediction information detection part in the Example of this invention. FIG. 3 is a block diagram illustrating a detailed configuration of a merge detection unit according to the first embodiment. 6 is a flowchart for explaining an operation of an encoded information deriving unit of the merge detecting unit of the first embodiment. 6 is a flowchart for explaining an operation of selecting neighboring blocks on the same picture in the encoding information deriving unit according to the first embodiment. FIGS. 14A to 14C are diagrams illustrating an operation of selecting a reference picture number candidate of a prediction block to be processed in the encoding information deriving unit according to the first embodiment. FIG. 6 is a diagram listing reference picture number selection patterns from reference picture number candidates of a prediction block to be processed in the encoding information deriving unit according to the first embodiment. 6 is a flowchart for explaining an operation of selecting coding information of a block adjacent to the same position of another picture having a different time in the coding information deriving unit according to the first embodiment. 7 is a flowchart for explaining detailed operation of calculating coding information of a block adjacent to the same position of another picture having a different time in the coding information deriving unit according to the first embodiment. It is a figure explaining the scaling which converts the distance of a col picture and its reference picture into the distance of the picture of a process target, and its reference picture, and matches a motion vector with the picture of a process target. 7 is a flowchart for explaining an operation of registering coding information of a reference adjacent block selected by a coding information deriving unit in a reference candidate list in a reference candidate list creating unit according to the first embodiment. In Example 1, it is a figure which shows an example of the reference candidate list produced by the reference candidate list production part. 7 is a flowchart for explaining an operation of updating the reference candidate list by detecting and deleting the same encoded information registered in the reference candidate list in the same information detecting unit of the first embodiment. FIG. 3 is a block diagram illustrating a detailed configuration of a merge determination unit according to the first embodiment. It is a block diagram which shows the detailed structure of the motion vector estimation part of Example 1. 6 is a flowchart for explaining the operation of the encoded information deriving unit of the motion vector predicting unit of the first embodiment. 6 is a flowchart for explaining an operation of selecting neighboring blocks on the same picture in the encoding information deriving unit according to the first embodiment. 6 is a flowchart of another method for explaining an operation of selecting neighboring blocks on the same picture in the encoding information deriving unit according to the first embodiment. It is a figure explaining scaling of a motion vector. In Example 1, it is a figure which shows an example of the reference candidate list produced by the reference candidate list production part. 7 is a flowchart for explaining an operation of detecting and deleting encoded information having the same motion vector registered in a reference candidate list and updating the reference candidate list in the same information detection unit according to the first embodiment. FIG. 3 is a block diagram illustrating a detailed configuration of a motion vector calculation unit according to the first embodiment. FIGS. 31A and 31B are diagrams illustrating the arrangement of adjacent blocks serving as reference destinations for the prediction block to be processed in the second embodiment. 12 is a flowchart for explaining an operation of an encoded information deriving unit of a merge detecting unit of the second embodiment. 12 is a flowchart for explaining an operation of selecting neighboring blocks on the same picture in the encoding information deriving unit according to the second embodiment. It is a figure explaining the operation | movement which selects the reference picture number candidate of the prediction block of a process target in the encoding information derivation part of Example 2. FIG. In the encoding information derivation part of Example 2, it is the figure which listed the selection pattern of the reference picture number from the reference picture number candidate of the prediction block of a process target. In Example 2, it is a figure which shows an example of the reference candidate list produced in the reference candidate list creation part of a merge detection part. It is a flowchart for demonstrating operation | movement of the encoding information derivation | leading-out part of the motion vector estimation part of Example 2. FIG. 12 is a flowchart for explaining an operation of selecting neighboring blocks on the same picture in the encoding information deriving unit according to the second embodiment. 10 is a flowchart of another method for explaining an operation of selecting neighboring blocks on the same picture in the encoding information deriving unit according to the second embodiment. In Example 2, it is a figure which shows an example of the reference candidate list produced in the reference candidate list creation part of a motion detection part. In Example 3, it is a figure which shows an example of the reference candidate list | wrist produced with the selection method different in merge mode and motion detection mode. It is a flowchart for demonstrating operation | movement of the encoding information derivation part of the merge detection part of Example 4. 14 is a flowchart for explaining an operation of selecting adjacent neighboring blocks on the same picture in the encoding information deriving unit according to the fourth embodiment. It is a flowchart for demonstrating operation | movement of the encoding information derivation | leading-out part of the motion vector prediction part of Example 4. 14 is a flowchart of a method for explaining an operation of selecting neighboring blocks on the same picture in the encoding information deriving unit according to the fourth embodiment. 14 is a flowchart for explaining an operation of registering, in a reference candidate list, encoding information of a reference adjacent block selected by an encoding information deriving unit in a reference candidate list creating unit of Example 5. In Example 6, it is a figure which shows an example of the reference candidate list produced in the reference candidate list production part. FIG. 20 is a block diagram illustrating a detailed configuration of a merge detection unit according to a seventh embodiment. FIG. 25 is a flowchart for explaining an operation of updating the reference candidate list by limiting the encoded information registered in the reference candidate list in the same information detection unit of the seventh embodiment. FIG. 20 is a block diagram illustrating a detailed configuration of a merge detection unit according to a seventh embodiment. FIG. 15 is a flowchart for explaining an operation of updating the reference candidate list by limiting the encoding information registered in the reference candidate list in the reference candidate list creating unit of the seventh embodiment. It is a block diagram which shows the detailed structure of the motion vector estimation part of Example 7. It is a flowchart for demonstrating the operation | movement which selects the surrounding adjacent block in the encoding information derivation | leading-out part of Example 7. It is a flowchart for demonstrating the detailed operation | movement of the calculation of the motion vector of the encoding information derivation part of Example 7. FIG. 25 is a flowchart for explaining an operation of registering in the reference candidate list the coding information of the reference adjacent block selected by the coding information deriving unit in the reference candidate list creating unit of the seventh embodiment. FIG. 25 is a flowchart for explaining an operation of updating the reference candidate list by limiting the encoded information registered in the reference candidate list in the same information detection unit of the seventh embodiment. It is a block diagram which shows the detailed structure of the motion vector estimation part of Example 7. FIG. 25 is a flowchart for explaining an operation of registering the coding information of the reference adjacent block selected by the coding information deriving unit in the reference candidate list in the reference candidate list creating unit of the eighth embodiment. In Example 8, it is a figure which shows an example of the reference candidate list | wrist produced in merge mode and motion detection mode. FIG. 25 is a flowchart for explaining an operation of registering, in a reference candidate list, coding information of reference neighboring blocks selected by a coding information deriving unit in a reference candidate list creation unit of Example 9. FIG. 25 is a flowchart for explaining an operation of creating / updating a conversion table in a reference candidate list creation unit according to the ninth embodiment. In the reference candidate list creation part of Example 9, it is a figure showing an example of correspondence of an index of a reference candidate list and an index of a conversion table. In Example 9, it is a figure which shows an example of the reference candidate list produced by the reference candidate list production part. It is a block diagram which shows the detailed structure of the motion vector estimation part of Example 12. 29 is a flowchart for explaining an operation of exchanging encoded information registered in a reference candidate list in a reference candidate control unit according to the twelfth embodiment.

  A preferred moving picture encoding apparatus and moving picture decoding apparatus for implementing the present invention will be described. FIG. 2 is a block diagram showing the configuration of the moving picture encoding apparatus 100 according to the embodiment of the present invention. The moving image encoding apparatus 100 includes an image memory 101, a motion vector detection unit 102, a motion vector prediction unit 103, an inter prediction information detection unit 104, a motion compensation prediction unit 105, a merge detection unit 106, a prediction method determination unit 107, and a switch 108. , First encoded bit string generation unit 109, residual signal generation unit 110, orthogonal transform / quantization unit 111, inverse quantization / inverse orthogonal transform unit 112, decoded image signal superimposition unit 113, encoded information storage memory 114, A decoded image memory 115, a second encoded bit string generation unit 116, and an encoded bit string multiplexing unit 117 are provided. The thick solid arrows connecting the blocks represent picture image signals, and the thin solid arrows represent the flow of parameter signals for controlling encoding.

  The image memory 101 temporarily stores image signals to be encoded supplied in the order of shooting / display time. The image memory 101 supplies the stored image signal to be encoded to the motion vector detection unit 102, the prediction method determination unit 107, and the residual signal generation unit 110 in units of predetermined pixel blocks. At this time, the images stored in the order of shooting / display time are rearranged in the encoding order and output from the image memory 101 in units of pixel blocks.

  The motion vector detection unit 102 calculates a motion vector for each prediction block size and each prediction mode by block matching or the like between the image signal supplied from the image memory 101 and the decoded image (reference picture) supplied from the decoded image memory 115. Detection is performed in units of prediction blocks, and the detected motion vectors are supplied to the motion compensation prediction unit 105, the motion vector prediction unit 103, and the prediction method determination unit 107. Here, the prediction modes are roughly classified as shown in FIG. 3A, and motion vectors in the unidirectional prediction or bi-prediction modes in FIGS. 3A and 3B are detected. The prediction mode will be described in the syntax definition described later.

  When the prediction mode is the motion detection mode, the motion vector prediction unit 103 uses the motion vector of the encoded information output from the inter prediction information detection unit 104 as a predicted motion vector, and the motion vector detected by the motion vector detection unit 102 A difference motion vector is calculated from the predicted motion vector, and the calculated difference motion vector is supplied to the prediction method determination unit 107. Further, an index for specifying the selected prediction motion vector is supplied to the prediction method determination unit 107. The detailed configuration and operation of the motion vector prediction unit 103 will be described later.

  The inter prediction information detection unit 104 predicts the processing target in the encoded prediction block adjacent to the periphery of the processing target prediction block on the same picture referred to by the processing target prediction block, or in another picture having a different time. The encoding information of the prediction block adjacent to the periphery of the block at the same position as the block is acquired from the encoding information storage memory 114. A plurality of references are obtained from the position information of the prediction block to be processed among the encoded information of neighboring neighboring blocks that are already encoded or stored in the encoded information storage memory 114 or adjacent blocks of different pictures of different times. The encoding information of the adjacent block that is the destination candidate is detected, and the encoding information of the adjacent block that is selected as the reference destination and the index that identifies the adjacent block are controlled by the moving image encoding apparatus 100 input to the switch 108. Switch in the prediction mode. The detailed configuration and operation of the inter prediction information detection unit 104 will be described later.

  The motion compensation prediction unit 105 uses the motion vector detected by the motion vector detection unit 102 or the coding information of the adjacent block selected as the reference destination detected by the inter prediction information detection unit 104 to perform motion compensation from the reference picture. A prediction image signal is generated by the prediction, and the prediction image signal is supplied to the prediction method determination unit 107. In the case of bi-prediction, L0 prediction, which is mainly used for forward prediction, and L1 prediction, which is mainly used for backward prediction, is adaptively multiplied by a weighting factor and superimposed. Then, a final predicted image signal is generated. This weighting coefficient is set in units of slices or prediction blocks.

  The merge detection unit 106 processes encoding information of an adjacent block (hereinafter referred to as a reference adjacent block) that is referred to by the processing target prediction block acquired by the inter prediction information detection unit 104 when the prediction mode is the merge mode. In order to use it as the encoding information of the target prediction block, the adjacent blocks are rearranged in the order of reference and registered in the reference candidate list. The merge detection unit 106 detects coding information registered in the reference candidate list as coding information of a prediction block to be processed, and detects a motion vector, a reference picture number, a reference list, and the like of the detected coding information Is supplied to the motion compensation prediction unit 105. Furthermore, the merge detection unit 106 supplies the prediction method determination unit 107 with an index that identifies the reference neighboring block including the detected coding information. The detailed configuration and operation of the merge detection unit 106 will be described later.

  The prediction method determination unit 107 evaluates the encoding information of the reference adjacent block, the code amount of the index for specifying the adjacent block, the distortion amount between the motion compensation prediction signal and the image signal, and the like. A prediction method including an optimal prediction block size, a partition mode (PartMode), a prediction mode (PredMode), and the like is determined. The prediction method determination unit 107 supplies the first encoded bit string generation unit 109 with information indicating the determined prediction method and encoded information including a difference motion vector or the like corresponding to the determined prediction method. Details of the prediction block size, the division mode, and the prediction mode will be described later.

  The switch 108 uses the motion vector prediction unit 103 or the merge information of the adjacent block selected as the reference destination detected by the inter prediction information detection unit 104 according to the prediction mode controlled by the video encoding device 100. It supplies by switching between detection parts 106. Further, the switch 108 stores the encoded information including information indicating the determined prediction method and a motion vector according to the determined prediction method in the encoded information storage memory 114, and corresponds to the determined prediction mode. The motion compensated prediction image signal is supplied to the residual signal generation unit 110 and the decoded image signal superimposition unit 113.

  The residual signal generation unit 110 generates a residual signal by subtracting the image signal to be encoded and the prediction signal and supplies the residual signal to the orthogonal transform / quantization unit 111. The orthogonal transform / quantization unit 111 performs orthogonal transform and quantization on the residual signal to generate an orthogonal transform / quantized residual signal, and the second encoded bit string generation unit 116 and the inverse quantization / quantization unit 116 This is supplied to the inverse orthogonal transform unit 112.

  The first encoded bit string generation unit 109 encodes the prediction method information determined by the prediction method determination unit 107 and the information on the difference motion vector according to the prediction method information according to a prescribed syntax rule, An encoded bit string is generated and supplied to the encoded bit string multiplexing unit 117.

  The weighting prediction weighting parameter supplied from the prediction method determination unit 107, the flag for identifying frame / field prediction at the time of interlace coding, the quantization quantization parameter, and the residual signal are encoded as necessary. A flag indicating whether or not, a flag for identifying an orthogonal transform method, a flag for identifying a coding order of residual signals, and information on a post filter such as a deblocking filter, and a predicted value of encoded information, which is actually used The difference from the value is supplied to the first encoded bit string generation unit 109 and encoded. Further, the predicted encoded information such as a flag for identifying the encoding order of the residual signal is stored in the encoded information storage memory 114.

  The second encoded bit string generation unit 116 generates a second encoded bit string by entropy encoding the orthogonally transformed and quantized residual signal according to a specified syntax rule, and the encoded bit string multiplexing unit 117 Supply. The encoded bit string multiplexing unit 117 multiplexes the first encoded bit string and the second encoded bit string in accordance with a specified syntax rule, and outputs a bit stream.

  The inverse quantization / inverse orthogonal transform unit 112 performs inverse quantization and inverse orthogonal transform on the orthogonal transform / quantized residual signal supplied from the orthogonal transform / quantization unit 111 to calculate a residual signal, and performs decoding. This is supplied to the image signal superimposing unit 113. The decoded image signal superimposing unit 113 superimposes the prediction signal according to the determination by the prediction method determining unit 107 and the residual signal that has been dequantized and inverse orthogonal transformed by the inverse quantization / inverse orthogonal transform unit 112 to decode the decoded image. Is generated and stored in the decoded image memory 115. The decoded image may be stored in the decoded image memory 115 after filtering processing for reducing distortion such as block distortion due to encoding. In that case, predicted encoded information such as a flag for identifying information of a post filter such as a deblocking filter is stored in the encoded information storage memory 114 as necessary.

  FIG. 4 is a block diagram showing a configuration of a moving picture decoding apparatus 200 in an embodiment corresponding to the moving picture encoding apparatus 100 of FIG. The moving picture decoding apparatus 200 includes a bit string separation unit 201, a first encoded bit string decoding unit 202, a second encoded bit string decoding unit 203, a motion vector calculation unit 204, an inter prediction information detection unit 205, and a merge determination unit 206. A motion compensation prediction unit 207, an inverse quantization / inverse orthogonal transform unit 208, a decoded image signal superimposing unit 209, an encoded information storage memory 210, a decoded image memory 211, and a switch 212. As in the moving picture encoding apparatus 100 of FIG. 2, the thick solid line arrows connecting the blocks represent picture image signals, and the thin solid line arrows represent the flow of parameter signals for controlling encoding.

  Since the decoding process of the moving picture decoding apparatus 200 in FIG. 4 corresponds to the decoding process provided in the moving picture encoding apparatus 100 in FIG. 2, the motion compensation prediction unit 207, the inverse quantum in FIG. 2, the decoded image signal superimposing unit 209, the encoded information storage memory 210, and the decoded image memory 211 are composed of the motion compensated prediction unit 105 and the inverse quantization of the moving image encoding apparatus 100 in FIG. 2. A function corresponding to each configuration of the inverse orthogonal transform unit 112, the decoded image signal superimposing unit 113, the encoded information storage memory 114, and the decoded image memory 115 is provided.

  The bit stream supplied to the bit string separation unit 201 is separated according to a rule of a prescribed syntax, and the separated encoded bit string is supplied to the first encoded bit string decoding unit 202 and the second encoded bit string decoding unit 203. Is done.

  The first encoded bit sequence decoding unit 202 decodes the supplied encoded bit sequence and outputs encoded information related to the prediction mode, motion vector, etc., and the encoded information is used as the motion vector calculation unit 204 or the inter prediction information detection unit. 205 and the motion compensation prediction unit 207 and also stored in the encoded information storage memory 210.

  The second encoded bit string decoding unit 203 calculates a residual signal that has been orthogonally transformed and quantized by decoding the supplied encoded bit string, and dequantized and quantized the residual signal that has been orthogonally transformed and quantized. This is given to the inverse orthogonal transform unit 208.

  When the prediction mode is the motion detection mode, the motion vector calculation unit 204 is determined by the inter prediction information detection unit 205, and the motion vector of the encoded information to be output is used as the prediction motion vector, and the first encoded bit string decoding unit A motion vector is calculated from the difference vector decoded in step 202 and the predicted motion vector, and the motion vector is supplied to the motion compensation prediction unit 207 and also supplied to the encoded information storage memory 210.

  The inter prediction information detection unit 205 is a decoded prediction block adjacent to the periphery of the prediction block to be processed on the same picture referenced by the prediction block to be processed, or a prediction block to be processed in another picture having a different time The encoding information of the prediction block adjacent to the periphery of the block at the same position is acquired from the encoding information storage memory 210. Among the encoded information of neighboring neighboring blocks that have already been decoded and are stored in the encoded information storage memory 210 or adjacent blocks of different pictures of different times, a plurality of reference destinations are obtained from the position information of the prediction block to be processed. Coding information of a candidate neighboring block is detected, and coding information of the neighboring block selected as a reference destination and an index specifying the neighboring block are decoded by the video decoding device 200 input to the switch 212. Supply by switching in mode. The detailed configuration and operation of the inter prediction information detection unit 205 will be described later.

  The merge determination unit 206 uses the encoding information of the processing target prediction block acquired by the inter prediction information detection unit 205 when the prediction mode is the merge mode as the encoding information of the processing target prediction block. In order to use it, the adjacent blocks are rearranged in the order of reference and registered in the reference candidate list. The merge determination unit 206 detects, from the reference candidate list, an adjacent block specified by an index that identifies the reference adjacent block decoded by the first encoded bit string decoding unit 202, and performs motion compensation on the encoded information. The data is supplied to the prediction unit 207 and also supplied to the encoded information storage memory 210. The detailed configuration and operation of the merge determination unit 206 will be described later.

  The motion compensation prediction unit 207 performs motion compensation prediction from the reference picture using the motion vector calculated by the motion vector calculation unit 204 or the coding information of the adjacent block selected as the reference destination detected by the merge determination unit 206. A predicted image signal is generated, and the predicted image signal is supplied to the decoded image signal superimposing unit 209. In the case of bi-prediction, the weighted coefficient is adaptively multiplied and superimposed on the two motion compensated prediction image signals of L0 prediction and L1 prediction to generate a final prediction image signal.

  The inverse quantization / inverse orthogonal transform unit 208 performs inverse orthogonal transform and inverse quantization on the orthogonal transform / quantized residual signal decoded by the first encoded bit string decoding unit 202, and performs inverse orthogonal transform. Obtain a dequantized residual signal.

  The decoded image signal superimposing unit 209 superimposes the predicted image signal subjected to motion compensation prediction by the motion compensation prediction unit 207 and the residual signal subjected to inverse orthogonal transform / inverse quantization by the inverse quantization / inverse orthogonal transform unit 208. As a result, the decoded image signal is decoded and stored in the decoded image memory 211. When stored in the decoded image memory 211, the decoded image may be stored in the decoded image memory 211 after filtering processing that reduces block distortion or the like due to encoding is performed on the decoded image.

  The switch 212 uses the motion vector calculation unit to calculate the encoding information of the adjacent block selected as the reference destination detected by the inter prediction information detection unit 205 according to the prediction mode decoded by the first encoded bit string decoding unit 202. 204 or the merge determination unit 206 is switched and supplied.

  In the following embodiments, a motion vector prediction method and a merge method that are commonly performed in the inter prediction information detection unit 104 of the video encoding device 100 and the inter prediction information detection unit 205 of the video decoding device 200 are used. Details of a method of integrating the arrangement of neighboring neighboring blocks to be referred to and selecting a reference neighboring block from them will be described.

  Before describing the method of selecting reference neighboring blocks in the embodiment of the present invention, terms used in the present embodiment are defined.

(Definition of terms used in the present invention)
Definition of coding block In this embodiment, as shown in FIG. 5, the screen is equally divided into square rectangular blocks of the same size. This block is called a coding block and is used as a basis for processing when performing coding and decoding. According to the texture in the screen, the coding block can be divided into four blocks with a small block size in order to optimize the coding process. An encoded block that is divided into equal screen sizes in the screen shown in FIG. 5 is called a maximum encoded block, and a block that is divided into four according to the encoding condition is collectively referred to as an encoded block. An encoded block having a minimum size that cannot be further divided into four is referred to as a minimum encoded block.

・ Definition of prediction block When performing motion compensation by dividing the screen into blocks, the smaller the motion compensation block size, the more detailed prediction can be performed. It adopts a mechanism to select the optimal one and perform motion compensation. A block that performs this motion compensation is called a prediction block. The prediction block is used when the inside of the coding block is divided, and according to motion compensation, the case where the inside of the coding block is regarded as one block without being divided is maximized, and is divided into two horizontally or vertically. And divided into four by vertical equal division. Depending on the division method, a division mode (PartMode) corresponding to the division type is defined and is shown in FIG. The numbers inside the rectangles in FIGS. 7A to 7D indicate the numbers of the divided prediction blocks. In order to manage the prediction block within the minimum coding block, numbers starting from 0 are assigned to the prediction block in order from top to bottom and from left to right.

  As described above, when motion detection / compensation is performed on an encoded block, the unit of motion detection / compensation includes the size of the encoded block itself, individual small blocks obtained by dividing the encoded block into two or four, and further 4 By dividing each divided small block in the same way as an encoded block, a block having a size obtained recursively up to the minimum size of the prediction block is obtained. In the following description, unless otherwise specified, the unit for motion detection / compensation is a “prediction block” regardless of the shape and size.

(Reference list)
Next, the reference list will be described. In encoding and decoding, a reference picture number is designated from a reference index for each reference list LX for reference. L0 and L1 are prepared, and 0 or 1 is entered in X. Inter prediction that refers to a reference picture registered in the reference list L0 is referred to as L0 prediction (Pred_L0), and motion compensation prediction that refers to a reference picture registered in the reference list L1 is referred to as L1 prediction (Pred_L1). L0 prediction is mainly used for forward prediction, L1 prediction is used for backward prediction, only L0 prediction is used for P slice, and L0 prediction, L1 prediction, and bi-prediction that averages or adds L1 prediction and L1 prediction are used for B slice. I can do it. In the subsequent processing, it is assumed that the value with the subscript LX in the output is processed for each L0 prediction and L1 prediction.

(Recording of encoding information)
Next, recording of encoded information such as motion vectors detected by motion detection / compensation will be described. Here, the coding information includes information used for coding, such as a motion vector, a reference list, a reference picture number indicating a reference picture registered in the reference list, and a prediction mode. Although motion detection / compensation is performed in units of prediction blocks, encoded information such as detected motion vectors is not recorded in units of prediction blocks, but is recorded in units of the minimum prediction block described above. That is, the same encoded information is recorded in a plurality of minimum prediction blocks constituting the prediction block.

  Since the size of the prediction block varies depending on the encoding condition, in order to record the prediction block unit in the encoding information storage memories 114 and 210, in addition to the detected motion vector, the position of the prediction block on the picture, the prediction block Additional information such as a shape is required, and access for obtaining encoded information is complicated. Therefore, although there is redundancy due to redundant recording of information, it is easier to access by dividing and recording into uniform minimum prediction block units.

  FIG. 8 shows an example in which a picture is divided into minimum prediction blocks, and the storage format of encoded information recorded in the encoded information storage memories 114 and 210 will be described with reference to this figure. In FIG. 8, w represents the width of the picture, and h represents the height of the picture. Assuming that the width and height of the minimum prediction block are p, a picture is composed of w / p minimum prediction blocks in the horizontal direction and h / p minimum prediction blocks in the vertical direction. Are recorded in the encoded information storage memories 114 and 210 in the raster scan order represented by the thick solid line in FIG.

  Here, a rectangular area surrounded by a thick line in FIG. 8 is a prediction block (an area indicated by hatching in FIG. 8). In the prediction block, encoded information such as a motion vector is detected by motion detection. The detected encoded information is recorded in the encoded information storage memories 114 and 210 corresponding to the minimum prediction block constituting the prediction block. Assuming that the top address of the encoded information recording area of the encoded information storage memories 114 and 210 is the origin in the upper left in FIG. 8, the start position of the prediction block is indicated by a black circle in FIG. Assuming that the start position of the prediction block is located at x positions to the right and y positions below the origin in the minimum prediction block unit, the start address of the prediction block is represented by w / p × y + x. The encoding information of the prediction block is recorded in the storage area in the encoding information storage memories 114 and 210 corresponding to the minimum prediction block in the prediction block indicated by hatching from the minimum prediction block at the head position. In this way, the same encoded information is recorded in the recording area of the encoded information storage memories 114 and 210 corresponding to the smallest prediction block in the prediction block.

  Next, the syntax of a bit stream of a moving image that is encoded by a moving image encoding device that includes a reference neighboring block selection method according to an embodiment of the present invention will be described.

(Syntax definition)
FIG. 3 (a) described above is one in which encoding is roughly classified according to the prediction mode. In the case of inter-image prediction (Inter mode), first, it is divided into a merge mode and a non-merge mode depending on whether or not the merge method is used. The non-merging mode is a mode for actually detecting a motion vector such as block matching, which can be called a motion detection mode. The merge mode is further divided into a skip mode and a non-skip mode. In the skip mode, the predicted image signal predicted by the coding information of the reference adjacent block is used as a decoded image, so that the image residual signal is zero. Therefore, the skip mode is a mode that does not require encoded transmission and encodes and transmits only an index representing a reference destination of encoded information. The coding information in the skip mode is the same as that in the merge mode, and here, the case where the image residual signal is encoded and transmitted is limited to the merge mode. FIG. 3B shows a table in which signals to be encoded in the Inter mode are classified. Unlike the other modes, the skip mode does not require coded transmission of an image residual signal. Therefore, the skip mode is determined before the determination in the order of branching of the prediction mode in FIG. This can reduce the code amount of the flag used for mode determination.

  FIG. 6 described above shows a syntax pattern described in units of prediction blocks in a slice. First, a flag skip_flag indicating whether or not the skip mode is set. When skip_flag is true (1) and the total number of reference adjacent blocks NumMergeCand exceeds 1, skip mode is applied, and the syntax element of the index of the merge list that is the candidate list of reference adjacent blocks is applied to the coding information. merge_idx is installed. When NumMergeCand is 1, one of them is a reference adjacent block, so that a reference adjacent block candidate is determined without transmitting merge_idx.

  Next, when the prediction mode of the prediction block is inter-picture prediction (Inter mode), a flag merge_flag indicating whether or not the mode is the merge mode is set. When merge_flag is true (1) and the total number of reference adjacent block candidates NumMergeCand exceeds 1, merge mode is applied, and the syntax information of the index of the merge list that is the candidate list of reference adjacent blocks is encoded information. merge_idx is installed. When NumMergeCand is 1, one of them is a reference adjacent block, so that a reference adjacent block candidate is determined without transmitting merge_idx.

  When merge_flag is false (0), the merge mode is not applied, normal motion vector detection is performed, and encoded information such as a motion vector and a reference picture number is transmitted (corresponding to the motion detection mode). For each prediction block, a reference list used for encoding and decoding is selected according to the slice type of the slice to which the prediction block belongs. When the slice type slice_type is “B”, a flag inter_pred_flag indicating unidirectional prediction or bi-prediction is set.

  inter_pred_flag is set to Pred_L0 (L0 prediction), Pred_L1 (L1 prediction) or Pred_BI (bi-prediction). Based on inter_pred_flag, a reference list LX (X = 0 or 1) is set. A syntax element ref_idx_lX representing a reference picture number obtained by motion vector detection for each set reference list, and a differential motion vector syntax element between the motion vector and a prediction motion vector of a neighboring block around the prediction block mvd_lX [i] is installed. Here, X represents a reference list with 0 or 1, i represents a differential motion vector component, i = 0 represents an x component, and i = 1 represents a y component.

  Next, when the total number of predicted motion vector candidates NumMvpCand (LX) exceeds 1, a syntax element mvp_idx_lX of an index of an MVP list that is a candidate list of predicted motion vectors to be referred to is set. NumMvpCand (LX) represents a function for calculating the total number of motion vector predictor candidates of the prediction block in the reference list LX (X is 0 or 1). Also, when the total number of predicted motion vector candidates NumMvpCand (LX) is one by the motion vector prediction method, the index mvp_idx_lX is set and is not encoded. This is because if the total number of motion vector predictor candidates is one, one of the motion vector predictors is a motion vector predictor, and therefore the motion vector predictor candidate to be referred to is determined without transmitting mvp_idx_lX. Based on the above syntax, it is possible to encode and decode a bit stream of a moving image.

  Next, FIG. 9A and FIG. 9B show an example of the arrangement of surrounding neighboring blocks having a predicted motion vector to be referred to in the motion detection mode or surrounding neighboring blocks serving as a reference destination of the encoded information in the merge mode. Show. In FIGS. 9A and 9B, adjacent blocks are represented by minimum prediction blocks with respect to the prediction block to be processed.

  This configuration accesses already encoded information recorded in the encoded information storage memories 114 and 210 based on the position and size (block width and height) of the prediction block to be processed. Convenient if you want to. Regarding the neighboring neighboring blocks shown in FIGS. 9A and 9B, the block adjacent to the left side of the prediction block to be processed is represented by the symbol “A”, and the block adjacent to the upper side is represented by “B”. The subscript number represents the position.

  The block “T” is an example of a block adjacent to the same position of another picture having a different time. In the embodiment of the present invention, the prediction block to be processed is set at the lower right of another picture. If the block T is outside the picture depending on the position of the prediction block to be processed, the position of the minimum prediction block near the center of another picture is set with respect to the prediction block to be processed. FIG. 9A shows adjacent blocks in a space on the same picture, which are five candidates, and are composed of six candidates, including candidates in different pictures with different times. FIG. 9B is an adjacent block in the space on the same picture, and is composed of five candidates, including four candidates and candidates in different pictures with different times.

  Although all adjacent blocks may be candidates as shown in FIG. 1, for example, the size of the prediction block to be processed is 64 × 64 pixels, and the encoded information is stored in the encoded information storage memories 114 and 210. Is 4 × 4 pixels, the number of blocks adjacent to the processing target prediction block is 16 on the left side, 16 on the upper side, and 3 blocks on the corner as the reference block. When implemented in hardware, it is necessary to consider the maximum number of memory accesses that can be assumed, the amount of memory, and the amount of processing (processing time).

  In the embodiment of the present invention, as shown in FIG. 9A, the candidates to be referred to are narrowed down by limiting the number of blocks adjacent to the left side and the block adjacent to the upper side to one, the diagonally lower left side, the right side. Blocks adjacent to the diagonally upper side and the diagonally left upper side are added as candidates. Further, in FIG. 9B, candidates to be referred to are narrowed down by limiting the blocks adjacent on the left side and the blocks adjacent on the upper side to one different from that in FIG. The upper adjacent block is a candidate.

  Thus, by limiting candidates in advance, the embodiment of the present invention has an effect of reducing the number of memory accesses, the amount of memory, and the amount of processing (processing time) without substantially reducing the coding efficiency. Further, in Embodiment 3 and later described later, the number of memory accesses, the amount of memory, and the amount of processing (processing time) can be reduced, and the code amount of the merge index and MVP index can be reduced. There is also an effect of reducing the amount. In the embodiment of the present invention, in the merge mode and the motion detection mode, the candidate positions are shared, thereby further reducing the number of memory accesses, the amount of memory, and the amount of processing (processing time).

  Hereinafter, description will be made using the arrangement example of FIG. When the arrangement example in FIG. 9B is used, the same processing as in the arrangement example in FIG. 9A can be performed by omitting C0 in the following description. The arrangement of the reference adjacent blocks may take any configuration as long as no contradiction occurs between the video encoding device 100 and the video decoding device 200. As long as the encoding information of the reference adjacent block can be stored in the encoding information storage memories 114 and 210 as information already encoded by the video encoding device 100 and already decoded by the video decoding device 200. Such information may be used.

  FIG. 10 shows details of the inter prediction information detection units 104 and 205 of the first embodiment installed in FIG. 2 showing the configuration of the video encoding device 100 and FIG. 4 showing the configuration of the video decoding device 200. Is. A portion surrounded by a thick dotted line in FIG. 10 indicates the inter prediction information detection units 104 and 205, which will be described with reference to FIG. The inter prediction information detection units 104 and 205 include a processing target block position detection unit 301, an adjacent block designation unit 302, and a temporary memory 303. In both the merge mode and the motion detection mode, the candidate positions are made common to the arrangement in FIG. 9A, and the encoding information of the adjacent blocks in the arrangement in FIG. 9A is encoded information storage memories 114 and 210. The procedure to acquire from will be described.

  First, information on the position and size of the prediction block to be processed in the picture is input to the processing block position detection unit 301. The position of the prediction block to be processed is represented by a pixel unit distance in the horizontal and vertical directions from the pixel at the upper left of the picture to the upper left pixel of the prediction block to be processed. Further, the width and height of the prediction block to be processed are calculated based on the PartMode representing the division mode of the encoding block to be processed and the number of times the encoding block has been recursively divided, and the prediction block to be processed refers to it. The position of the reference adjacent block is calculated.

  Based on the calculated position of the adjacent block, the adjacent block designating unit 302 accesses the encoded information storage memories 114 and 210, and encodes corresponding to the position of the adjacent block shown in FIG. Information is read into the temporary memory 303. The temporary memory 303 is a memory that is provided inside the inter prediction information detection units 104 and 205 and stores the coding information of the reference neighboring block that is referred to by the prediction block to be processed. Here, when the calculated position of the adjacent block is outside the screen, since the adjacent block does not exist, “0” is recorded in the temporary memory 303.

  As described above, by sharing the candidate position in both the merge mode and the motion detection mode, the amount of processing for storing the encoded information of the adjacent block in the temporary memory 303 is reduced on the encoding side. Further, it is not necessary to access the encoding information storage memory 114 in order to change the prediction method for each prediction mode and acquire the encoding information of the adjacent blocks necessary each time. On the other hand, since the decoding side can start the process of storing the coding information of the adjacent block in the temporary memory 303 before performing the discrimination between the merge mode and the motion detection mode, the speed of the decoding process can be improved. I can do it.

  In the moving picture coding apparatus 100 and the moving picture decoding apparatus 200 including the coding information selection method according to the embodiment of the present invention, coding information such as a motion vector and a reference picture number of a prediction block to be processed is directly coded. Instead, the encoding information of neighboring neighboring blocks already encoded / decoded stored in the encoding information storage memories 114 and 210 is referred to and used. Then, by encoding the index indicating the position of the reference adjacent block and the index of the motion vector predictor that predicts the motion vector, the amount of code related to the encoded information can be reduced. In the following, a method of deriving an index indicating a position of a reference adjacent block and an index of a motion vector predicting a motion vector in an encoding process from the encoding information of the adjacent block acquired by the inter prediction information detection units 104 and 205 And a specific example of a method for obtaining encoding information based on an index indicating a position of a reference adjacent block decoded in a decoding process corresponding to the encoding process and an index of a motion vector predicting a motion vector Will be shown.

Example 1
First, the merge detection unit 106 according to the first embodiment installed in FIG. 2 showing the configuration of the video encoding device 100 will be described. FIG. 11 shows a detailed configuration of the merge detection unit 106. A portion surrounded by a thick dotted line in FIG. 11 indicates the merge detection unit 106, which will be described with reference to FIG. The merge detection unit 106 includes an encoded information derivation unit 310, a reference candidate list creation unit 311, an identical information detection unit 312, an output unit 313, and a reference candidate list storage memory 314.

  First, the output destination of the inter prediction information detection unit 104 is switched to the merge detection unit 106 according to the prediction mode controlled by the moving image encoding apparatus 100 input to the switch 108. Coding information of the reference adjacent block that is referred to by the prediction block to be processed, which is stored in the temporary memory 303 inside the inter prediction information detection unit 104, is input to the coding information deriving unit 310.

  The operation of the encoded information deriving unit 310 will be described using the flowchart of FIG. It is determined whether or not the coding information of the adjacent block in the same picture as the prediction block to be processed is valid from among the adjacent blocks input first (S100).

  FIG. 13 shows a detailed flow of determination processing for each adjacent block in the same picture as the prediction block to be processed. First, the variable N is initialized (S200). In the variable N, adjacent blocks A1, B1, B0, A0, C0 on the same picture shown in FIG. 9A are set. Here, N = A1 is set at the time of initialization, and the variable N is updated in the order of B1, B0, A0, C0.

Next, a variable for storing coding information of adjacent blocks used in the subsequent determination is initialized. Here, the variables are a flag availableFlagN indicating whether the adjacent block is valid, a motion vector mvLXN, a reference picture number refIdxLXN, and a flag predFlagLXN indicating the validity of the reference list, and are initialized as follows (S201).
availableFlagN = 0
mvLXN = (0,0)
refIdxLXN = 0
predFlagLXN = 0
Here, the subscript X is set to 0 or 1 representing the reference list. The position of the adjacent block of variable N (hereinafter referred to as adjacent block N) and encoding information are acquired (S202).

  Based on the acquired position of the adjacent block N, it is determined whether the adjacent block N is valid (S203). For example, when the prediction block to be processed is located at the left end of the picture, there is no adjacent block to the left of the prediction block to be processed, so there is no corresponding encoding information in the encoding information storage memory 114. It is determined to be invalid. When the adjacent block N is invalid (N in S203), the available FlagN is set to “0” (S207). When the adjacent block N is valid (Y in S203), it is determined whether or not the prediction mode of the adjacent block N is an intra-frame coding (Intra) mode (S204).

  If the prediction mode of the adjacent block N is Intra mode (Y in S204), availableFlagN is set to “0” (S207). When the prediction mode of the adjacent block N is not Intra mode (N in S204), availableFlagN is set to “1” (S205). Subsequently, the coding information of the adjacent block N is set by substituting it into refIdxLXN, mvLXN, and predFlagLXN (S206).

  When the determination process for the adjacent block N is completed as described above, it is determined whether or not the variable N is the last of the adjacent blocks (S208). Since the variable N is updated in the order of A1, B1, B0, A0, and C0, it is determined here whether N is C0 or not. If N is C0 (Y in S208), it means that all adjacent blocks have been determined, and the process ends. If N is not C0 (N in S208), N is updated (S209). N is updated in the order of the above-mentioned adjacent blocks, and the processing after step S201 is repeated for the adjacent block N. As described above, the encoding information is derived using the neighboring neighboring blocks existing on the same picture as the prediction block to be processed as the reference block.

  Subsequently, returning to FIG. 12, the reference picture number of the prediction block to be processed when another picture having a different time is set as a merge candidate is determined (S101). This process is performed in order to determine a reference picture used when the prediction block to be processed is inter-predicted by referring to the motion vector of a block adjacent to the same position of another picture having a different time, which will be described next. At this time, the reference picture number of the picture referred to by the prediction block to be processed is determined from the coding information of the already-encoded adjacent block. Here, a method based on coding information of adjacent blocks on the same picture derived in the previous process (S100) will be described with reference to FIG.

  First, the four adjacent blocks derived in step S100 are classified into three groups. As a group, an adjacent block in contact with the left side of the prediction block to be processed is group A, an adjacent block in contact with the upper side is group B, and an adjacent block in contact with a corner is group C.

  First, the group A will be described with reference to FIG. Since the adjacent block belonging to the group A is A1, the flag availableFlagA1 indicating whether or not A1 in step S100 is valid is checked (S300). If availableFlagA1 is 1 (Y in S300), the group A reference picture number refIdxLXA is set to the reference picture number of A1. Otherwise (N in S300), refIdxLXA is set to -1 indicating that there is no reference picture number.

  Similarly, the group B will be described with reference to FIG. Since the adjacent block belonging to the group B is B1, the flag availableFlagB1 indicating whether or not B1 in step S100 is valid is checked (S301). If availableFlagB1 is 1 (Y in S301), the group B reference picture number refIdxLXB is set to the reference picture number of B1. Otherwise (N in S301), refIdxLXB is set to -1.

  The adjacent blocks belonging to group C are A0, B0, and C0, and determination is performed in the order of B0, A0, and C0 as shown in FIG. Similar to A1 and B1, the flag availableFlagB0 of B0 is examined (S302). If availableFlagB0 is 1 (Y in S302), the reference picture number refIdxLXC of group C is set to the reference picture number of B0, otherwise (N in S302), the flag availableFlagA0 of A0 is examined (S303). If availableFlagA0 is 1 (Y in S303), the reference picture number refIdxLXC of group C is set to the reference picture number of A0. Otherwise (N in S303), the C0 flag availableFlagC0 is checked (S304). If availableFlagC0 is 1 (Y in S304), the reference picture number refIdxLXC of group C is set to the reference picture number of C0. Otherwise (N in S304), refIdxLXC is set to -1.

  Note that, when there is no block at the position of C0 as in the adjacent block on the same picture shown in FIG. 9B, the determination is made for the two adjacent blocks of B0 and A0. By selecting the majority or minimum value of the reference picture numbers refIdxLXN (N is A, B or C) of each group derived in this way, the reference picture numbers of blocks adjacent to the same position of another picture having different times are determined. .

  FIG. 15 shows determination based on the reference picture number refIdxLXN. Description will be made in order from the top of FIG. The first to fourth lines are cases where the same value exists in refIdxLXN, and is selected by majority vote. The first line is the case where refIdxLXN is all the same. If it is -1, 0 is set, otherwise refIdxLXA is set. The 2nd to 4th lines are cases where two of the three refIdxLXNs are the same. If the value when the two are the same is -1, the value of the remaining one refIdxLXN is the value when the other two are the same.

  The fifth to eighth lines are cases where the same value does not exist in refIdxLXN, and the minimum value is selected from them. The value of the smallest reference picture number is selected except for refIdxLXN, which is -1 among the three. As described above, the reference picture number of the prediction block to be processed is determined, but methods other than this method may be used.

  Returning to FIG. 12, next, it is determined whether or not the coding information of the block adjacent to the same position of another picture having a different time is valid (S102). FIG. 16 shows a detailed flow of determination processing for adjacent blocks in a different picture from the prediction block to be processed. First, another picture having a different time from the picture including the prediction block to be processed is designated (S400). When the slice including the prediction block to be processed is a P slice, only the L0 prediction can be used, so the picture indicated by the reference picture number 0 in the L0 reference list is designated. In the case of a B slice, one of L0 prediction and L1 prediction and L0 and L1 bi-prediction can be used, but the picture indicated by the reference picture number 0 in the reference list of either L0 or L1 is designated. Here, the designation of another picture having a different time in the case of the B slice is not mentioned, but for example, a parameter for determining which reference list is referred to in the case of the B slice is inserted in the slice header. This method is also conceivable.

  Subsequently, the position of a block adjacent to the same position of another picture with a different reference time is designated (S401). In the first embodiment, since the reference adjacent block “T” shown in FIG. 9A is a block adjacent to the same position of another picture having a different time, PartMode representing the position and division mode of the prediction block to be processed. Based on the number of divisions, the width and height of the prediction block to be processed are calculated, and the position of the adjacent block “T” is calculated. When the calculated position of the adjacent block is outside the picture, it is set to a position near the center of the same position as the prediction block to be processed in another picture having a different time. In the following description, another picture having a different time is a col picture, a block adjacent to the same position of a processing target block of the picture is a col block, and “Col” is added as a subscript to the motion vector or reference picture number. It shall be expressed with an addition.

  Based on the position of the col picture and the block col block adjacent to the same position as the prediction block to be processed in the col picture, the encoded information of the col block is read by accessing the encoded information storage memory 114, and the encoded information is valid. Whether or not, and a motion vector is derived (S402).

  FIG. 17 shows the detailed processing of those derivations and will be described with reference to this figure. It is determined whether or not the col block is valid from the coding information and position of the col block read from the coding information storage memory 114 (S500). If the col block does not exist or the prediction mode of the col block is Intra (Y in S500), the reference motion vector mvCol of the col block is set to (0, 0), the reference valid flag availableFlagCol is set to 0 (S501), step The process proceeds to S509. If not (N in S500), the process proceeds to determination of the flag predFlagL0 indicating the validity of the L0 reference list of the col block (S502).

  When predFlagL0 is 0 (Y in S502), since L0 prediction is invalid, encoding information for L1 prediction is selected, mvCol is set to mvL1, reference reference picture number refIdxCol is set to refIdxL1 (S503), and the process proceeds to step S507. If not (N in S502), that is, if the L0 prediction is valid, the process proceeds to the determination of the flag predFlagL1 indicating the validity of the L1 reference list of the col block (S504).

  When predFlagL1 is 0 (Y in S504), since L1 prediction is invalid, encoding information for L0 prediction is selected, mvCol is set to mvL0, refIdxCol is set to refIdxL0 (S505), and the process proceeds to step S507. If not (N in S504), that is, if both the L0 and L1 predictions are valid, the process proceeds to derivation in the case of the Pred_BI mode (S506). In derivation in the case of the Pred_BI mode, either L0 prediction or L1 prediction is selected.

  The selection method is, for example, selecting the same reference list as the col picture selected in step S500, selecting the reference picture of the col block L0 prediction and L1 prediction that has a shorter inter-image distance from the col picture, or For example, it is possible to select a direction in which the motion vectors of L0 prediction and L1 prediction of the col block intersect with the processing target picture.

  An example of the third criterion is shown in FIG. In this example, when the picture to be processed intersects the motion vector of the col block, that is, the reference picture of the col picture, the picture to be processed, and the col picture are displayed in time order, and the block corresponding to the col block is always processed. When evaluating the motion of the surrounding area within a very short time, it is more reliable to select a crossing-side prediction to obtain a closer value in motion vector prediction. . Therefore, the reference list side of the motion vector that intersects with the picture to be processed is selected, and mvCol and refIdxCol are set.

  Returning to FIG. 17, next, since the motion vector exists, the reference valid flag availableFlagCol is set to 1 (S507). The variable X indicating the reference list is set to 0 (S508), and the motion vectors for L0 and L1 prediction of the col block are derived. The position of the reference picture is specified from the standard motion vector mvCol and the standard reference picture number refIdxCol obtained in the above process, and the inter-image distance colDist with the col picture is calculated.

  On the other hand, the position of the reference picture is specified from the reference picture number refIdxLX of the processing target prediction block determined in step S401, and the inter-image distance currDist with the processing target picture is calculated. The currDist and the colDist are compared (S509). When currDist and colDist are the same (OK in S509), the reference motion vector mvCol is set as mvLXCol as it is (S510).

・・・（式１） When currDist and colDist are different (NG in S509), distance conversion by scaling is performed. In scaling, the motion of the prediction block to be processed is calculated by converting the motion vector of the adjacent block to a distance by matching the inter-image distance of the picture referenced by the adjacent block with the inter-image distance from the picture to which the prediction block to be processed refers. This is a vector processing. FIG. 18 shows an example of scaling. Since the inter-image distance between the col block on the col picture and the picture to which the prediction block to be processed is referred is colDist and currDist, the inter-image distance of the prediction block to be processed is calculated using the following equation for the motion vector mvCol of the col block. Accordingly, a motion vector is calculated (S511).

... (Formula 1)

  It is determined whether X is 1 (S512). When X is not 1, that is, 0 (N in S512), the L1 prediction direction is calculated in the same procedure. Therefore, X is updated to 1 (S513), and step S509 and subsequent steps are repeated. If X is 1 (Y in S512), the process ends.

  The encoding information of the adjacent blocks obtained in this way is input to the reference candidate list creation unit 311. In the first embodiment, a reference candidate list is provided in the reference candidate list storage memory 314 as a storage area for registering encoding information candidates of reference adjacent blocks. Then, priorities are assigned to the encoding information candidates of the reference adjacent block, and the encoding information candidates of the reference adjacent block are registered in the reference candidate list in descending order of priority. As a result, the code amount of the index merge_idx in the reference candidate list can be reduced.

  The code amount is reduced by arranging the element having a higher position of each element in the reference candidate list in front of the reference candidate list. An index merge_idx indicating priority is assigned in ascending order from 0. For example, when there are three elements in the reference candidate list, index 0 of the reference candidate list is “0” (binary notation), index 1 is “10” (binary notation), and index 2 is “11” (binary notation). The code amount representing index 0 is 1 bit, and the code amount is reduced by registering an element considered to have a high occurrence frequency in index 0.

  The reference candidate list provided in the reference candidate list storage memory 314 has a list structure, and stores an index indicating the location in the reference candidate list and encoding information candidates of reference adjacent blocks corresponding to the index as elements. An area is provided. This arrangement region is represented by candList. The index number starts from 0, and encoding information candidates of reference adjacent blocks are stored in the storage area of the reference candidate list candList. In the subsequent processing, the encoding information of the index i registered in the reference candidate list candList is represented by candList [i], and is distinguished from the reference candidate list candList by array notation. Note that the encoding information stored in the storage area of the reference candidate list is represented by the position name (A0, A1, B0, B1, C0, T) of the reference adjacent block unless otherwise specified.

  The operation of the reference candidate list creation unit 311 will be described with reference to the flowchart of FIG. First, the variable N and the index k of the reference candidate list are initialized (S600). In the variable N, the adjacent block A1 shown in FIG. 9A is initialized, and k is set to 0. The index k indicates the priority order of the storage areas for the encoded information candidates set in the storage area of the reference candidate list.

  Next, the valid flag availableFlagN of the reference adjacent block N is determined (S601). When availableFlagN is 1 (Y in S601), the encoding information of the adjacent block N is registered in the reference candidate list candList [k] (S602), and k is updated (S603). If availableFlagN is 0 (N in S601), it is not registered in the reference candidate list, and proceeds to the next.

  It is determined whether or not the adjacent block N is the last reference block (S604). If it is the last block (Y in S604), the index value is set to the total number of candidate lists NumListCand (S605), and the process ends. If it is not the last block (N in S604), the variable N is updated (S606), and the processes after step S601 are repeated.

  Here, the order of adjacent blocks in which the variable N is updated is the order of priority for storage in the reference candidate list. In the first embodiment, in the merge mode, the order is (A1, B1, B0, A0, C0, T). Is set. A code having a shorter code length is assigned to a merge index for specifying a merge candidate in the reference candidate list, as the higher priority is recorded in front of the reference candidate list.

  In the merge mode, priority is given to adjacent blocks A1 and B1 in which the side of the prediction block to be processed is in contact with the side of the prediction block to be processed that is most likely to have the same encoding information. Register in front of the candidate list. This reduces the code amount of the merge index and improves the encoding efficiency.

  With the above processing, if all the reference adjacent blocks registered in the reference candidate list are valid, the reference candidate list is created in the order shown in FIG. Using the priority as an index, each codeword is represented by the right column of the reference candidate list, and the maximum codeword length is NumListCand-1. In the case of FIG. 20, NumListCand is 6, and the code word is represented by a maximum of 5 bits. When only one reference adjacent block is valid, the maximum codeword length is 0. Therefore, no index is required, and the candidate of encoding information of the adjacent block determined to be valid is unique as a reference destination. Will be decided.

  The created reference candidate list is input to the same information detection unit 312. The same information detection unit 312 compares the encoded information candidates stored in the reference candidate list, and if there is the same encoded information candidate, the encoded information having the smallest reference candidate list index is stored. All but the candidates are deleted. The operation will be described with reference to the flowchart shown in FIG.

  First, variables n and m representing indexes of the reference candidate list are set to 0 and 1, respectively (S700). The encoded information of indexes n and m stored in the reference candidate list is compared (S701). Here, the encoding information to be compared is the prediction mode, the reference list being used, the reference picture number for each reference list being used, and the motion vector for each reference list being used. If it is determined that there is a mismatch (N in S701), the process proceeds to step S704. If it is determined that they match (Y in S701), it is determined whether n and m have the larger index, that is, whether m is already recorded in the deletion list (S702).

  If m is already recorded in the deletion list (Y in S702), the process proceeds to step S704. If not recorded yet (N in S702), m is recorded in the deletion list (S703). The deletion list is a temporary storage memory provided in the same information detection unit 312.

  Next, 1 is added to m and updated (S704). m is compared with the total number of candidate lists NumListCand (S705). When m is not NumListCand (N in S705), the comparison with the encoded information of the index n after step S701 is repeated. When m becomes NumListCand (Y in S705), 1 is added to n and updated (S706).

  Next, n is compared with (NumListCand-1) (S707). When n is not (NumListCand-1) (N in S707), m is set to (n + 1) (S709), and the comparison of the encoded information after step S701 is repeated. When n becomes (NumListCand-1) (Y in S707), the encoded information in the storage area of the list corresponding to the index recorded in the deletion list is deleted, and the candidate with a small index with reference to index 0 The total number NumListCand of the codeword and reference candidate list is updated (S708), and the process ends.

  Finally, the output unit 313 outputs the index and encoding information in the created reference candidate list. The reference candidate list outputs a reference candidate list as a merge list and an index in the list as a merge index. In the moving image encoding apparatus 100, the video is output to the motion compensation prediction unit 105 and the prediction method determination unit 107.

  The merge determination unit 206 according to the first embodiment installed in the video decoding device 200 corresponding to the merge detection unit 106 according to the first embodiment installed in the above-described video encoding device 100 will be described. FIG. 22 shows a detailed configuration of the merge determination unit 206. A portion surrounded by a thick dotted line in FIG. 22 indicates the merge determination unit 206. The merge determination unit 206 includes an encoded information derivation unit 310, a reference candidate list creation unit 311, an identical information detection unit 312, a reference candidate list storage memory 314, and a selection unit.

  The merge determination unit 206 corresponds to the merge detection unit 106 of the video encoding device 100, and the internal configuration is different only in the selection unit 315 and the output unit 313. Other encoding information deriving units 310 and reference candidates are different. The list creation unit 311, the same information detection unit 312, and the reference candidate list storage memory 314 have the same functions. Accordingly, the same reference candidate list as that in the encoding process is created by the processing up to the same information detection unit 312.

  Below, the selection part 315 which acquires the encoding information of the reference adjacent block in merge mode from the produced reference candidate list is demonstrated. The selection unit 315 selects an adjacent block in the reference candidate list specified by an index for specifying the reference adjacent block decoded by the first encoded bit string decoding unit 202 from the created reference candidate list. The selected encoded information is supplied to the motion compensation prediction unit 207 and also supplied to the encoded information storage memory 210.

  Next, the motion vector predicting unit 103 according to the first embodiment installed in FIG. 2 illustrating the configuration of the moving image encoding device 100 will be described. FIG. 23 shows a detailed configuration of the motion vector prediction unit 103. A portion surrounded by a thick dotted line in FIG. 23 indicates the motion vector predicting unit 103, which will be described with reference to FIG. The motion vector prediction unit 103 includes an encoded information derivation unit 320, a reference candidate list creation unit 321, an identical information detection unit 322, an output unit 323, a reference candidate list storage memory 324, and a difference motion vector calculation unit 326.

  Coding information of the reference adjacent block that is referred to by the prediction block to be processed stored in the temporary memory 303 inside the inter prediction information detection unit 104 is input to the coding information deriving unit 320. The operation of the encoded information deriving unit 320 will be described with reference to the flowchart of FIG.

  It is determined whether or not the coding information of the neighboring block in the same picture as the processing target prediction block is valid from among the neighboring blocks input first (S800). FIG. 25 shows a detailed flow of determination processing for each adjacent block in the same picture as the prediction block to be processed. Note that the determination process of FIG. 25 is performed independently for each list of L0 and L1.

  Depending on the prediction mode of the prediction block to be processed, only the coding information registered in the reference list L0 is registered in the case of L0 prediction in unidirectional prediction, and is registered in the reference list L1 in the case of L1 prediction by unidirectional prediction. In the case of bi-prediction, determination processing of each encoded information registered in the reference lists L0 and L1 is performed. In this specification, unless otherwise noted, the determination process in the motion detection mode will be described with one reference candidate list as LX, and only one reference candidate list shown in the figure will be given.

  First, the variable N is initialized (S900). In the variable N, adjacent blocks A1, B1, B0, A0, C0 on the same picture shown in FIG. 9A are set. Here, N = A1 is set at the time of initialization, and the variable N is updated in the order of B1, B0, A0, C0.

Next, a variable for storing coding information of adjacent blocks used in the subsequent determination is initialized. Here, the variables are a flag availableFlagLXN indicating whether or not the adjacent block is valid, a motion vector mvLXN, a reference picture number refIdxLXN, and a flag predFlagLXN indicating the validity of the reference list, and are initialized as follows (S901).
availableFlagLXN = 0
mvLXN = (0,0)
refIdxLXN = 0
predFlagLXN = 0
Here, the subscript X is set to 0 or 1 representing the reference list. The position of the adjacent block of variable N (hereinafter referred to as adjacent block N) and the encoding information are acquired (S902).

  Based on the acquired position of the adjacent block N, it is determined whether or not the adjacent block N is valid (S903). For example, when the prediction block to be processed is located at the left end of the picture, there is no adjacent block to the left of the prediction block to be processed, so there is no corresponding encoding information in the encoding information storage memory 114. It is determined to be invalid.

  When the adjacent block N is invalid (N in S903), the availableFlagLXN is set to “0” (S908). When the adjacent block N is valid (Y in S903), it is determined whether or not the prediction mode of the adjacent block N is an intra-frame coding (Intra) mode (S904).

  When the prediction mode of the adjacent block N is Intra mode (Y in S904), the availableFlagLXN is set to “0” (S908). If the prediction mode of the adjacent block N is not Intra mode (N in S904), whether or not the prediction using the same reference picture number in the same reference list as that of the calculation target prediction of the processing target prediction block is also performed in the adjacent block N Is determined (S905).

  If not performed (N in S905), the availableFlagLXN is set to “0” (S908). If it has been performed (Y in S905), the availableFlagLXN is set to "1" (S906). Subsequently, the value of the motion vector of the adjacent block N is substituted into mvLXN and set (S907).

  When the determination process for the adjacent block N is completed as described above, it is determined whether or not the variable N is the last of the adjacent blocks (S909). Since the variable N is updated in the order of A1, B1, B0, A0, and C0, it is determined here whether N is C0 or not. If N is C0 (Y in S909), all adjacent blocks have been determined, and the process ends. If N is not C0 (N in S909), N is updated (S910). N is updated in the order of the adjacent blocks described above, and the processes in and after step S901 are repeated for the adjacent block N.

  In the first embodiment, it is determined whether or not the prediction using the same reference picture number in the same reference list as that of the calculation target prediction of the processing target prediction block is performed in the adjacent block N in step S905. This is because, if the conditions are the same reference picture number and reference list, the motion vector difference value is high because there is a high probability that the motion vector of the adjacent prediction block to be processed and the adjacent block N will be the same or close to each other. Becomes smaller, and the coding efficiency is improved.

However, there is not always an adjacent block that performs prediction using the same reference picture number and reference list on the same picture. If there is not, a motion vector difference value cannot be calculated, and coding efficiency Will be reduced. Therefore, whether or not the adjacent block is valid even when the prediction performed in the adjacent block satisfies the following condition without narrowing the determination under the condition of the same reference list and reference picture number in step S905. The flag availableFlagLXN is set to 1 (valid), and the coding information is derived as a reference adjacent block candidate.
Condition 1: Same reference picture, same reference picture number Condition 2: Different reference list, same reference picture number Condition 3: Same reference list, different reference picture number Condition 4: Different reference list, different reference picture number

  The processing flow in this case is shown in FIG. FIG. 26 differs from FIG. 25 in that the condition determination of only condition 1 in step S905 of FIG. 25 is deleted, the condition determination of conditions 1 to 4 in step S911 is performed, and if condition 1 or 2 is not satisfied, the movement The vector scaling process (S912) is added. Since other steps S900 to S910 are the same as those in FIG. 25, the changed / added portions will be described below.

  If the prediction mode of the adjacent block N is not Intra mode (N in S904), the availableFlagLXN is set to “1” (S906). Subsequently, conditions 1 to 4 are determined by comparing the encoding information of the prediction target of the calculation target prediction block with the encoding information of the adjacent block N (S911). When condition 1 or condition 2 is satisfied, the value of the motion vector of the same reference list (when condition 1 is met) or a different reference list (when condition 2 is met) of adjacent block N is substituted into mvLXN and set (S907).

  When the condition 3 or the condition 4 is satisfied, the scaling processing of the motion vector of the adjacent block N in the same reference list (when the condition 3 is met) or different reference list (when the condition 4 is met) is performed (S912). In the motion detection mode, motion vector prediction is performed in order to transmit the motion vector of the prediction block to be processed with a smaller code amount. Motion vector prediction is a process of taking a difference between a motion vector of a prediction block to be processed and a motion vector of an adjacent block and encoding the difference value. At this time, if the picture to be referred to is different, it affects the size of the motion vector, and the difference value may be larger than the motion vector of the actual prediction block to be processed.

  Therefore, scaling is performed in which the motion vector of the adjacent block is converted into a distance by matching the inter-image distance of the picture referenced by the adjacent block with the inter-image distance from the picture referenced by the processing target prediction block. FIG. 27 shows an example of scaling. The motion vector detection unit 102 performs motion vector detection on the processing target prediction block, and detects the motion vector, reference picture number, and reference list of the processing target prediction block.

・・・（式２） On the other hand, since the adjacent block on the processing target picture has its own coding information, a picture to be referred to is selected from each reference picture number. The inter-image distances td and tb between the adjacent block on the processing target picture and the picture referred to by the processing target prediction block are calculated, and the motion vector mvLX of the same reference list LX of the adjacent block N (when condition 3 is met) ) Or a motion vector mvLY (when condition 4 is met) of a different reference list LY (Y = 1 when X = 0, Y = 0 when X = 1) between the images of the prediction block to be processed by the following equation: The distance is converted to match the distance.

... (Formula 2)

  Returning to FIG. 26, the converted motion vector mvLX or mvLY is substituted for mvLXN and set (S907). By performing this scaling process, it is possible to derive a closer predicted motion vector for the prediction block to be processed and improve accuracy. As described above, the encoding information is derived using the neighboring neighboring blocks existing on the same picture as the prediction block to be processed as the reference block.

  Returning to FIG. 24, next, it is determined whether or not the coding information of the block adjacent to the same position of another picture having a different time is valid (S801). This process is basically the same as the process of step S103 of the encoded information deriving unit 310 in the merge detection unit 106. In the merge mode, the encoding information deriving unit 310 in the merge detecting unit 106 derives the reference picture number of the prediction block to be processed (S102) before performing the process of step S103.

  In contrast, in the motion detection mode, the motion vector detection unit 102 performs motion vector detection on the processing target prediction block, and detects the motion vector, reference picture number, and reference list of the processing target prediction block. Is done. A picture to be referred to from each reference picture number is selected from the coding information of the detected block to be processed and the coding information of a block adjacent to the same position of another picture whose known time is different, and processed. A motion vector of a block adjacent to the same position of another picture having a different time is converted into an inter-image distance between the target prediction block and its reference picture.

  The encoding information of the adjacent block obtained in this way is input to the reference candidate list creation unit 321. The reference candidate list created by the reference candidate list creation unit 321 is created for each of the two reference lists for L0 prediction and L1 prediction, according to the Inter mode of the prediction block to be processed. In the Inter mode, L0 prediction and L1 prediction are defined as unidirectional prediction, and Bi-pred prediction is defined as bi-prediction, and two reference lists of L0 prediction and L1 prediction are used in Bi-pred prediction. Accordingly, one or two reference candidate lists may be provided in the reference candidate list storage memory 314 according to the prediction mode at the time of encoding, or reference candidate lists for two L0 predictions and two L1 predictions may be provided in advance. It may be provided. In this specification, unless otherwise noted, only one reference candidate list in the motion detection mode will be described for convenience of explanation, and only one reference candidate list shown in the figure will be given.

  In the first embodiment, a reference candidate list is provided in the reference candidate list storage memory 324 as a storage area for registering encoding information candidates of reference neighboring blocks. Priorities of encoding information candidates of the reference adjacent block are given priority, and encoding information candidates of the reference adjacent block are registered in the reference candidate list storage memory 324 in descending order of priority. Thereby, it is possible to reduce the code amount of the indexes mvp_idx_l0 and mvp_idx_l1 of the reference candidate list.

  The amount of codes is reduced by placing elements with higher priorities in front of the reference candidate list. An index indicating the position of each element in the reference candidate list is assigned in ascending order from 0. The reference candidate list provided in the reference candidate list storage memory 324 has a list structure, and stores an index indicating the location inside the reference candidate list and encoding information candidates of reference adjacent blocks corresponding to the index as elements. An area is provided.

  This arrangement region is represented by candListLX. The number of the index starts from 0, and the encoding information candidates of the reference adjacent block are stored in the storage area of the reference candidate list candListLX. In the subsequent processing, the encoding information of the index i registered in the reference candidate list candListLX is represented by candListLX [i], and is distinguished from the reference candidate list candListLX by array notation. Note that the encoding information stored in the storage area of the reference candidate list is represented by the position name (A0, A1, B0, B1, C0, T) of the reference adjacent block unless otherwise specified.

  The operation of the reference candidate list creation unit 321 is equivalent to the reference candidate list creation unit 311 of the merge detection unit 106, and a flag availableFlagLXN indicating whether or not the adjacent block obtained by the encoding information deriving unit 320 is valid, Determination is made in the order A1, B1, A0, B0, C0, T defined in the motion detection mode, and if it is 1 (valid), it is registered in the reference candidate list provided in the reference candidate list storage memory 324.

  With the above processing, if all of the reference adjacent blocks are registered in the reference candidate list and are valid, the reference candidate list is created in the order shown in FIG. Using the priority as an index, each codeword is represented by the right column of the reference candidate list, and the maximum codeword length is NumListCand-1. In the case of FIG. 20, NumListCand is 6, and the code word is represented by a maximum of 5 bits. When only one reference adjacent block is valid, the maximum codeword length is 0. Therefore, no codeword is required, and the candidate of encoding information of an adjacent block determined to be valid is unique as a reference destination. Will be determined.

  The created reference candidate list is input to the same information detection unit 322. In the same information detection unit 322, when motion vectors stored in the reference candidate list are compared and there is a candidate having the same motion vector, all of them are deleted except for the candidate having the index of the smallest reference candidate list. The The operation will be described with reference to the flowchart shown in FIG.

  First, variables n and m representing indexes of the reference candidate list are set to 0 and 1 respectively (S1000). The motion vectors of indexes n and m stored in the reference candidate list are compared (S1001). If it is determined that there is a mismatch (N in S1001), the process proceeds to step S1004. If it is determined that they match (Y in S1001), it is determined whether n and m have the larger index, that is, whether m is already recorded in the deletion list (S1002).

  If m is already recorded in the deletion list (Y in S1002), the process proceeds to step S1004. If not recorded yet (N in S1002), m is recorded in the deletion list (S1003). The deletion list is a temporary storage memory provided in the same information detection unit 322.

  Next, 1 is added to m and updated (S1004). m is compared with the total number of candidate lists NumListCand (S1005). When m is not NumListCand (N of S1005), the comparison with the motion vector of the index n after step S1001 is repeated. When m becomes NumListCand (Y in S1005), 1 is added to n and updated (S1006).

  Next, n is compared with (NumListCand-1) (S1007). If n is not (NumListCand-1) (N in S1007), m is set to (n + 1) (S1009), and the comparison of motion vectors after step S1001 is repeated. When n becomes (NumListCand-1) (Y in S1007), the encoded information in the storage area of the list corresponding to the index recorded in the deletion list is deleted, and the candidate with a small index with reference to index 0 The code word and the total number of reference candidate lists NumListCand are updated (S1008), and the process ends.

  The difference motion vector calculation unit 326 calculates the difference motion vector from the motion vector detected by the motion vector detection unit 102 and the predicted motion vector, using the motion vector of the encoded information in the reference candidate list thus created as the prediction motion vector. Then, the calculated difference motion vector is supplied to the output unit 323.

  Finally, the output unit 323 outputs the index and the difference motion vector in the created reference candidate list. The reference candidate list outputs the reference candidate list as an MVP list and the index in the list as an MVP index. In the moving image encoding apparatus 100, the image is output to the prediction method determination unit 107.

  The motion vector calculation unit 204 according to the first embodiment installed in the video decoding device 200 corresponding to the motion vector prediction unit 103 according to the first embodiment installed in the above-described video encoding device 100 will be described. . FIG. 30 shows a detailed configuration of the motion vector calculation unit 204. A portion surrounded by a thick dotted line in FIG. 30 indicates the motion vector calculation unit 204. The motion vector calculation unit 204 includes an encoded information derivation unit 320, a reference candidate list creation unit 321, an identical information detection unit 322, a reference candidate list storage memory 324, a selection unit 325, and a motion vector addition unit 327.

  The motion vector calculation unit 204 corresponds to the motion vector prediction unit 103 of the video encoding device 100, and the internal configuration is different only in the selection unit 325, the output unit 323, and the motion vector addition unit 327. The encoded information deriving unit 320, the reference candidate list creating unit 321, the same information detecting unit 322, and the reference candidate list storage memory 324 have the same functions. Accordingly, the same reference candidate list as that in the encoding process is created by the processing up to the same information detection unit 322.

  Below, the selection part 325 which acquires the encoding information of the reference adjacent block in motion detection mode from the produced reference candidate list is demonstrated. The selection unit 325 selects an adjacent block in the reference candidate list specified by an index for specifying the reference adjacent block decoded by the first encoded bit string decoding unit 202 from the created reference candidate list. A motion vector is output as a predicted motion vector from the encoded information of adjacent blocks in the selected list, and the motion vector adder 327 predicts the motion vector and the first encoded bit string decoder 202 decodes the motion vector. Are added to calculate a motion vector. The motion vector is supplied to the motion compensation prediction unit 207, and the encoding information of the adjacent block of the selected reference candidate list is supplied to the encoding information storage memory 210.

(Example 2)
31A and 31B show the arrangement of reference adjacent blocks with respect to the processing target prediction block in the second embodiment. The symbols of the adjacent blocks are the same as those in FIGS. 9A and 9B, and the description will be made using the arrangement example in FIG. 31A. In the second embodiment, adjacent blocks on the same picture as the prediction block to be processed are divided into a block group adjacent to the left side (A0 and A1 in FIG. 31A), and a block group adjacent to the upper side (FIG. 31A ) Are grouped into B0, B1, and C0), and one reference adjacent block is selected from each block group as a representative value of the block group.

  As a selection method, it is the same as in the first and second embodiments to determine whether an adjacent block is valid based on the prediction mode and position of the encoding information of the adjacent block, but one block group at a time. The difference is that representative neighboring blocks are selected. In this way, a method of selecting an adjacent block representing the block group from a block group composed of a plurality of adjacent blocks is called a scan. By this scan, the reference candidate list includes three adjacent blocks: an adjacent block representing the block group adjacent to the left side, an adjacent block representing the block group adjacent to the upper side, and a block adjacent to the same position of another picture at different times. Will be registered.

  Accordingly, the total number of adjacent blocks registered in the reference candidate list can be reduced as compared with the first embodiment, the number of comparisons of encoded information in the same information detection unit 312, and the motion vector in the same information detection unit 322. The number of comparisons can be reduced. Furthermore, the codeword length assigned to the index of the reference candidate list is shortened, and the code amount of the index of the reference candidate list can be reduced. Furthermore, by changing the processing order of adjacent blocks in the block group adjacent to the left and upper sides according to the prediction mode, it is possible to select an adjacent block suitable for the prediction mode, and an improvement in coding efficiency can be expected. .

  In the processing order, when the prediction mode is the merge mode, the merge detection unit 106 and the merge determination unit 206 are arranged such that the block group adjacent to the left side as shown by the thin dotted arrows in FIGS. The block group adjacent to is processed in the order of B1, B0, C0 with the upper left adjacent block C0 as the last order. In the motion detection mode, the motion vector predicting unit 103 and the motion vector calculating unit 204 are adjacent to the block group adjacent to the left side from the bottom to the top side as indicated by the thin solid arrows in FIGS. 31 (a) and 31 (b). Blocks are processed from right to left.

  In the merge mode, coding efficiency is improved by preferentially determining and selecting adjacent blocks A1 and B1, which are considered to be most likely to have the same encoding information as the prediction block to be processed. On the other hand, in the motion detection mode, in order to widen the selection range of the predicted motion vector, the distance between the left and upper candidates is set so that a difference in encoding information is easily generated between the left and upper adjacent block candidates. The adjacent blocks A0 and B0 that are far away are preferentially determined and selected. Thereby, the code amount of a difference motion vector is reduced and encoding efficiency is improved.

  Hereinafter, only the operation of the second embodiment different from the first embodiment will be described in detail. First, the operation of the merge detection unit 106 in the video encoding apparatus 100 will be described. The merge detection unit 106 has the same configuration as that of FIG. 11 described in the first embodiment, but the processing of the encoded information deriving unit 310 is different from that of the first embodiment. explain. FIG. 32 is a flowchart illustrating the operation of the encoded information deriving unit 310 according to the second embodiment.

  First, the output destination of the inter prediction information detection unit 104 is switched to the merge detection unit 106 in accordance with the prediction mode controlled by the video encoding device 100 input to the switch 108, and the internal prediction information detection unit 104 has an internal destination. The coding information of the reference neighboring block that is stored in the temporary memory 303 and is referred to by the prediction block to be processed is input to the coding information deriving unit 310.

  It is determined whether or not the coding information of the adjacent block in the same picture as the prediction block to be processed is valid from the input adjacent blocks (S103). FIG. 33 shows a detailed flow of determination processing for each adjacent block in the same picture as the prediction block to be processed. 13 is a process in which step S210, step S211, step S212, step S213, and step S214 are newly added to the flowchart shown in FIG. 13 of the first embodiment, and the other processes are the same as in the first embodiment. Here, only the added process will be described in detail.

  First, the block group M is initialized (S210). Here, the block group adjacent to the left side is set, and M, which is a value indicating the left side block group, is set. Next, the variable N in the block group is initialized (S200). The variable N is set according to the processing order in the block group described above, and when N is updated, the left block group is updated in the order of A1, A0, and the upper side is updated in the order of B1, B0, C0.

  The processing from step S201 to step S209 in FIG. 33 is basically the same as that in FIG. 13 of the first embodiment, but it is determined whether or not the prediction mode of the adjacent block N is the intra-frame coding (Intra) mode ( S204) When the prediction mode of the adjacent block N is not the Intra mode (N in S204), the subsequent processing is different from that of the first embodiment.

  It is determined whether or not the prediction mode of the adjacent block N is the intra-frame coding (Intra) mode (S204). If the prediction mode of the adjacent block N is not the Intra mode (N in S204), whether or not M is the left block group. (S211), if M is the left block group (Y in S211), the availableFlagM is set to "1" (S205), and the encoding information of the adjacent block N is set to refIdxLXM, mvLXM, and predFlagLXM. Substitute and set (S206). Only when M is the upper block group (N in S211), it is determined whether or not the encoding information of the adjacent block selected in the left block group and the encoding information of the adjacent block N are the same ( S212).

  Since an adjacent block representing the left block group is first selected, the encoding information is retained and used for determination of the upper block group. By this processing, since it does not overlap with the coding information of the adjacent block representing the left block group, the range of options as a reference destination is expanded. In order to reduce the amount of processing, the processing in steps S211 and S212 can be omitted and the processing can proceed directly to step S205.

  When the encoding information of the adjacent block selected in the left block group is different from the encoding information of the adjacent block N or there is no encoding information of the adjacent block selected in the left block group (N in S212), availableFlagM Is set to “1” (S205), and the encoding information of the adjacent block N is substituted into refIdxLXM, mvLXM, and predFlagLXM (S206).

  Next, after the coding information of the block group M is set, whether or not the variable M is a value indicating the upper block group in order to perform processing on the next block group, that is, the block group adjacent on the upper side. (S213). If M is not B (upper side) (N in S213), M is updated to B (upper side) (S214), and the processes after step S200 are repeated. If M is on the upper side (Y in S213), the process is terminated.

  When the adjacent block N is not valid (N of S203), when the prediction mode of the adjacent block N is not the intra-frame coding (Intra) mode (Y of S204), the encoding information of the adjacent block selected in the left block group If the encoded information of the adjacent block N is the same (Y in S212), availableFlagM is set to "0" (S207), and it is determined whether or not the adjacent block N is the last of the adjacent blocks in the block group M. (S208).

  The update of the variable N is set according to the processing order in the block group described above. Here, it is determined whether N is A0 (in the case of the left block group) or C0 (in the case of the upper block group). If the adjacent block N is the last block (Y in S208), the process proceeds to the determination of the block group M (S213). Otherwise (N in S208), the variable N is updated according to the processing order in the block group. (S209), the process after step S201 is repeated. As described above, the encoding information is derived as the reference block representing the adjacent block of the block group existing on the same picture as the prediction block to be processed.

  Returning to FIG. 32, subsequently, the reference picture number of the prediction block to be processed when another picture having a different time is set as the merge candidate is determined (S104). Although the process of step S101 in the first embodiment may be used as it is, here, the encoding information of the block group adjacent to the left side and the upper side on the same picture derived in the previous process (S103) is used. A technique based on this will be described with reference to FIGS.

  First, the flag availableFlagM indicating whether or not the adjacent blocks of the block group A adjacent to the left side and the block group B adjacent to the upper side derived in step S103 are valid and the reference picture number refIdxLXM are input. Here, M is A and B.

  First, the left adjacent block group A will be described with reference to FIG. A flag availableFlagA indicating whether the adjacent block in the left adjacent block group is valid is checked (S1100). If availableFlagA is 1 (Y in S1100), the reference picture number refIdxLXA of the adjacent block in the left adjacent block group is set. Otherwise (N in S1100), refIdxLXA is set to -1.

  Similarly, the upper adjacent block group B will be described with reference to FIG. A flag availableFlagB indicating whether or not the adjacent block of the upper adjacent block group is valid is checked (S1101). If availableFlagB is 1 (Y in S1101), the reference picture number refIdxLXB of the adjacent block in the upper adjacent block group is set. Otherwise (N in S1101), refIdxLXB is set to -1.

  Based on the condition of the reference picture number refIdxLXM (M is A, B) of each block group thus derived, the reference picture number refIdxLXCol of a block adjacent to the same position of another picture having a different time is determined. FIG. 35 shows determination based on the reference picture number refIdxLXM. The first line in FIG. 35 is a case where refIdxLXA and refIdxLXB are the same value and not −1, and refIdxLXA is selected. The second line is a case where refIdxLXA and refIdxLXB are different values, both of which are not -1, and the value of the smallest reference picture number is selected. Either one of the third and fourth lines is -1, and the one that is not -1 is selected. The fifth line is a case where both are −1. At this time, “0” is set. As described above, the reference picture number of the prediction block to be processed is determined, but methods other than this method may be used.

  Returning to FIG. 32, next, it is determined whether or not the coding information of the block adjacent to the same position of another picture having a different time is valid (S102). Since this is the same as step S102 in the first embodiment, a description thereof will be omitted. The encoding information of the adjacent blocks obtained in this way is input to the reference candidate list creation unit 311.

  Also in the second embodiment, as in the first embodiment, a reference candidate list is provided in the reference candidate list storage memory 314 as a storage area for registering encoding information candidates of reference adjacent blocks, and the reference adjacent block encoding information candidates are set as candidates. Priorities are assigned, and encoding information candidates of reference neighboring blocks are registered in the reference candidate list in descending order of priority. Thereby, the code amount of the index merge_idx in the reference candidate list is reduced.

  The reference candidate list creation unit 311 performs the same operation as in the first embodiment, and the left and upper adjacent block groups to be referred to, the valid flag availableFlagM (where M is A , B, T) is determined, and the encoding information of adjacent blocks is registered in the reference candidate list candList. Here, the order of adjacent blocks in which the variable N is updated is the priority order for storing in the reference candidate list. In the second embodiment, the same order (A, B, T) at different times is set on the left side, upper side. I will do it.

  With the above processing, if all the reference adjacent blocks registered in the reference candidate list are valid, the reference candidate list is created in the order shown in FIG. Using the priority as an index, each codeword is represented by the right column of the reference candidate list, and the maximum codeword length is 2. Here, the parentheses in the reference candidate list represent one adjacent block that is processed by the encoding information deriving unit 310 in the order from left to right in the parentheses from the left or upper adjacent block group.

  In the created reference candidate list, the same information detection unit 312 deletes the encoded information stored in the reference candidate list that becomes the same encoded information, and the reference candidate list created by the reference candidate list creating unit 311 Output the index and encoding information. The reference candidate list is output with the reference candidate list as a merge list and an index in the list as a merge index. In the moving image encoding apparatus 100, the video is output to the motion compensation prediction unit 105 and the prediction method determination unit 107.

  The merge determination unit 206 according to the second embodiment installed in the video decoding device 200 corresponding to the merge detection unit 106 according to the second embodiment installed in the above-described video encoding device 100 will be described. The merge determination unit 206 has the same configuration as that of FIG. 22 described in the first embodiment, and the processing of the encoded information deriving unit 310 is different from that of the first embodiment. Other reference candidate list creation unit 311, identical information detection unit 312, reference candidate list storage memory 314, and selection unit 315 have the same functions as those in the first embodiment.

  Further, since the encoding information deriving unit 310 of the merge determining unit 206 in the second embodiment has the same function as the encoding information deriving unit 310 of the merge detecting unit 106 in the second embodiment described above, the merge information in the second embodiment is used. The same reference candidate list as that of the merge detection unit 106 according to the second embodiment is created by the processing up to the same information detection unit 312 of the merge determination unit 206 of the determination unit 206. The selection unit 315 that acquires the encoding information of the reference adjacent block in the merge mode from the created reference candidate list will be described. The selection unit 315 selects an adjacent block in the reference candidate list specified by an index for specifying the reference adjacent block, decoded by the first encoded bit string decoding unit 202, from the created reference candidate list. The selected encoded information is supplied to the motion compensation prediction unit 207 and also supplied to the encoded information storage memory 210.

  Next, the motion vector prediction unit 103 according to the second embodiment installed in FIG. 2 illustrating the configuration of the moving image encoding device 100 will be described. The motion vector predicting unit 103 according to the second embodiment has the same configuration as that of FIG. 23 described in the first embodiment, but the processing of the encoded information deriving unit 320 is different from that of the first embodiment. The operation of the unit 320 will be described. FIG. 37 is a flowchart showing the operation of the encoded information deriving unit 320 in the second embodiment.

  First, the output destination of the inter prediction information detection unit 104 is switched to the motion vector prediction unit 103 according to the prediction mode controlled by the video encoding device 100 input to the switch 108. Coding information of the processing target prediction block stored in the temporary memory 303 inside the inter prediction information detection unit 104 and the reference adjacent block is input to the coding information deriving unit 320. It is determined whether or not the coding information of the adjacent block in the same picture as the prediction block to be processed is valid from the input adjacent blocks (S802). FIG. 38 shows a detailed flow of determination processing for each adjacent block in the same picture as the prediction block to be processed.

  25 is a process in which step S913, step S914, step S915, step S916, and step S917 are newly added to the flowchart shown in FIG. 25 of the first embodiment, and the other processes are the same as in the first embodiment. Here, only the added process will be described in detail.

  First, the variable M of the block group is initialized (S913). Here, the block group adjacent to the left side is set, and M, which is a value indicating the left side block group, is set. Next, the variable N in the block group is initialized (S900). The variable N is set according to the processing order in the block group described above, and when N is updated, the left block group is updated in the order of A0, A1, and the upper side is updated in the order of B0, B1, C0. The subsequent processing from step S901 to step S910 is basically the same as in the first embodiment, but it is determined whether or not the prediction mode of the adjacent block N is the intra-frame coding (Intra) mode (S904). When the N prediction mode is not the Intra mode (N in S904), the subsequent processing is different from that in the first embodiment.

  It is determined whether or not the prediction mode of the adjacent block N is the intra-frame coding (Intra) mode (S904). If the prediction mode of the adjacent block N is not the Intra mode (N in S904), whether or not M is the left block group. (S914), and if M is the left block group (Y in S914), the process proceeds to step S905. Only when M is the upper block group (N in S914), it is determined whether or not the motion vector mvLXA of the adjacent block selected in the left block group and the motion vector of the adjacent block N are the same (S915). ).

  Since an adjacent block representing the left block group is selected first, the motion vector is retained and used for determination of the upper block group. This process eliminates the overlap with the motion vector of the adjacent block representing the left block group, so that the range of options as a reference destination is expanded. In order to reduce the amount of processing, the processing in steps S914 and S915 can be omitted and the processing can proceed directly to step S905.

  When the motion vector mvLXA of the adjacent block selected in the left block group and the motion vector of the adjacent block N are the same (Y in S915), the process proceeds to step S908. If they are different (N in S915), it is determined whether prediction using the same reference picture number in the same reference list as the calculation target prediction of the processing target prediction block is performed in the adjacent block N (S905).

  When prediction using the same reference picture number is performed in the adjacent reference block N in the same reference list as the prediction using the motion vector that is the calculation target of the processing target prediction block, the availableFlagLXN is set to “1”. Setting is performed (S906), and the motion vector of the adjacent block N is substituted into mvLXM and set (S907). After the motion vector of the block group M is set, it is determined whether or not the variable M is a value B indicating the upper block group in order to perform processing on the next block group, that is, the block group adjacent on the upper side ( S916). If M is not B (upper side) (N in S916), M is updated to B (upper side) (S917), and the processes in and after step S900 are repeated. If M is B (upper side) (Y in S916), the process is terminated.

  When the adjacent block N is not valid (N in S903), when the prediction mode of the adjacent block N is not the intra-frame coding (Intra) mode (Y in S904), the motion vector mvLXA of the adjacent block selected in the left block group If the motion vector of the adjacent block N in the upper block group is the same (Y in S915), if prediction using the same reference picture number is not performed in the same reference list (N in S905), the availableFlagLXM is set to “0”. (S908), it is determined whether or not the adjacent block N is the last of the adjacent blocks in the block group M (S909). The update of the variable N is set according to the processing order in the block group described above. Here, it is determined whether N is A1 (in the case of the left block group) or C0 (in the case of the upper block group).

  If the adjacent block N is the last block (Y in S909), the process proceeds to determination of the block group M (S916). Otherwise (N in S909), the variable N is updated in accordance with the processing order in the block group (S910), and the processes in and after step S901 are repeated. As described above, the encoding information is derived as the reference block representing the adjacent block of the block group existing on the same picture as the prediction block to be processed.

In the second embodiment, as in the first embodiment, even when the prediction performed in the neighboring block satisfies the following condition, the flag availableFlagLXM indicating whether the neighboring block is valid is set to 1 (valid), and the reference neighboring The encoding information can be derived as a block candidate.
Condition 1: Same reference picture, same reference picture number Condition 2: Different reference list, same reference picture number Condition 3: Same reference list, different reference picture number Condition 4: Different reference list, different reference picture number

  FIG. 39 shows the flow of processing in this case. The difference from FIG. 38 described above is that the condition determination of only condition 1 in step S905 is deleted, the condition determination of conditions 1 to 4 in step S911 is performed, and if condition 1 or 2 is not satisfied, motion vector scaling processing is performed (S912) is added.

  First, the variable M of the block group is initialized (S913). Here, the block group adjacent to the left side is set, and M is set to A. Next, the variable N in the block group is initialized (S900). The variable N is set according to the processing order in the block group described above, and when N is updated, the left block group is updated in the order of A0, A1, and the upper side is updated in the order of B0, B1, C0. The subsequent processing from step S901 to step S910 is basically the same as that in FIG. It is determined whether or not the prediction mode of the adjacent block N is the intra-frame coding (Intra) mode (S904). If the prediction mode of the adjacent block N is not the Intra mode (N in S904), the subsequent processing is shown in FIG. Different from 38.

  When the prediction mode of the adjacent block N is not the Intra mode (N in S904), that is, in the case of the Inter mode, the coding information of the prediction target of the calculation target prediction block and the coding information of the adjacent block N are compared. Then, the determination process of conditions 1 to 4 is performed (S911). When the condition 1 or 2 is satisfied, the process proceeds to a determination process of whether or not M is the left block group (S914). On the other hand, when the condition 3 or 4 is satisfied, the scaling process of the motion vector of the adjacent block N is performed in the same manner as S912 in FIG. 26 of the first embodiment (S912), and the determination process whether or not M is the left block group (S914).

  When M is the left block group (Y in S914), availableFlagLXM is set to “1” (S906). If M is the upper block group (N in S914), it is determined whether or not the motion vector mvLXA of the adjacent block selected in the left block group and the motion vector of the adjacent block N are the same (S915). In order to reduce the amount of processing, the processing in step S914 and step S915 can be omitted and the processing can proceed directly to step S906.

  When the motion vector mvLXA of the adjacent block selected in the left block group and the encoding information of the adjacent block N are the same (Y in S915), the availableFlagLXN is set to “0” (S908), and when it is different (N in S915) , AvailableFlagLXM is set to “1” (S906).

  Similarly to step S907 of FIG. 26 in the first embodiment, the motion vector mvLX of the same reference list (when the condition 1 is met) of the adjacent block N or the motion vector mvLY of the different reference list (when the condition 2 is met), or the same The motion vector mvLX converted by scaling of the reference list (when condition 3 is met) or the motion vector mvLY converted by scaling of a different reference list (when condition 4 is matched) is substituted into mvLXM and set ( S907).

  After the motion vector of the block group M is set, it is determined whether or not the variable M is the upper side in order to perform processing for the next block group, that is, the block group adjacent to the upper side (S916). If M is not on the upper side (N in S916), M is updated to B (upper side) (S917), and the processes after step S900 are repeated. If M is on the upper side (Y in S916), the process is terminated.

  When availableFlagLXM is set to “0” (S908), it is determined whether or not the adjacent block N is the last of the adjacent blocks in the block group M (S909). The update of the variable N is set according to the processing order in the block group described above. Here, it is determined whether N is A1 (in the case of the left block group) or C0 (in the case of the upper block group). If the adjacent block N is the last block (Y in S909), the process proceeds to the determination of the block group M (S916). Otherwise (Y in S909), the variable N is updated according to the processing order in the block group. (S910), the process after step S901 is repeated. As described above, the motion vector is derived as a reference block representing an adjacent block of a block group existing on the same picture as the prediction block to be processed.

  Returning to FIG. 37, next, it is determined whether or not the coding information of the block adjacent to the same position of another picture having a different time is valid (S801). Since this is the same as step S801 in the first embodiment, a description thereof will be omitted. The encoding information of the adjacent block obtained in this way is input to the reference candidate list creation unit 321. The reference candidate list created by the reference candidate list creating unit 321 is created for each of the two reference lists of L0 prediction and L1 prediction. In this specification, unless otherwise noted, only one reference candidate list in the motion detection mode will be described for convenience of explanation, and only one reference candidate list shown in the figure will be given.

  In the second embodiment, a reference candidate list is provided in the reference candidate list storage memory 324 as a storage area for registering encoding information candidates of reference neighboring blocks. Priorities of encoding information candidates of reference adjacent blocks are prioritized, and encoding information candidates of reference adjacent blocks are registered in the reference candidate list storage memory 324 in descending order of priority. Thereby, it is possible to reduce the code amount of the indexes mvp_idx_l0 and mvp_idx_l1 of the reference candidate list.

  The amount of codes is reduced by placing elements with higher priorities in front of the reference candidate list. The index indicating the priority order is assigned in ascending order from 0. The reference candidate list provided in the reference candidate list storage memory 324 has a list structure, and stores an index indicating the location inside the reference candidate list and encoding information candidates of reference adjacent blocks corresponding to the index as elements. An area is provided. This arrangement region is represented by candListLX. The number of the index starts from 0, and the encoding information candidates of the reference adjacent block are stored in the storage area of the reference candidate list candListLX. In the subsequent processing, the encoding information of the index i registered in the reference candidate list candListLX is represented by candListLX [i], and is distinguished from the reference candidate list candListLX by array notation. Note that the encoding information stored in the storage area of the reference candidate list is represented by the position name (A0, A1, B0, B1, C0, T) of the reference adjacent block unless otherwise specified.

  The reference candidate list creation unit 321 performs the same operation as that in the first embodiment, and the left and upper adjacent block groups to be referred to, the valid flag availableFlagLXM of the block adjacent to the same position of another picture with different time (where M is A , B, T) is registered, and the encoding information of adjacent blocks is registered in the reference candidate list candListLX. Here, the order of adjacent blocks in which the variable N is updated is the priority order for storing in the reference candidate list. In the second embodiment, the same order (A, B, T) at different times is set on the left side, upper side. I will do it. With the above processing, if all the reference adjacent blocks registered in the reference candidate list are valid, the reference candidate list is created in the order shown in FIG.

  Using the priority as an index, each codeword is represented by the right column of the reference candidate list, and the maximum codeword length is 2. Here, the parentheses in the reference candidate list represent one adjacent block that is processed by the encoding information deriving unit 320 in the order from the left to the right in the parentheses from the left or upper adjacent block group. When only one reference adjacent block is valid, the maximum codeword length is 0. Therefore, no codeword is required, and the candidate of encoding information of an adjacent block determined to be valid is unique as a reference destination. Will be determined.

  In the created reference candidate list, the same information detection unit 322 deletes the encoded information stored in the reference candidate list that is the same encoded information, and the reference candidate list created by the difference motion vector calculation unit 326 The motion vector of the encoded information in the image is used as a predicted motion vector, a difference motion vector is calculated from the motion vector detected by the motion vector detection unit 102 and the predicted motion vector, and the calculated difference motion vector is supplied to the output unit 323. .

  Finally, the output unit 323 outputs the index and the difference motion vector in the created reference candidate list. The reference candidate list outputs the reference candidate list as an MVP list and the index in the list as an MVP index. In the moving image encoding apparatus 100, the image is output to the prediction method determination unit 107.

  The motion vector calculation unit 204 of Example 2 installed in the video decoding device 200 corresponding to the motion vector prediction unit 103 of Example 2 installed in the above-described video encoding device 100 will be described. . The motion vector calculation unit 204 has the same configuration as that in FIG. 30, and the processing of the encoded information deriving unit 320 is different from that in the first embodiment. Other reference candidate list creation unit 321, identical information detection unit 322, reference candidate list storage memory 324, selection unit 325, and motion vector addition unit 327 have the same functions as in the first embodiment.

  Also, since the encoded information deriving unit 320 has the same function as the encoded information deriving unit 320 of the motion vector predicting unit 103 in the second embodiment described above, the steps up to the same information detecting unit 322 of the motion vector calculating unit 204 are provided. In the process, the same reference candidate list as that of the motion vector prediction unit 103 of the second embodiment is created. The selection unit 325 that acquires the coding information of the reference adjacent block in the motion detection mode from the created reference candidate list will be described. The selection unit 325 selects an adjacent block in the reference candidate list specified by an index for specifying the reference adjacent block decoded by the first encoded bit string decoding unit 202 from the created reference candidate list. The motion vector is output as the predicted motion vector from the coding information of the adjacent block of the selected reference candidate list, and the motion vector adding unit 327 decodes the predicted motion vector and the first coded bit string decoding unit 202. The motion vector is calculated by adding the vectors, supplied to the motion compensation prediction unit 207, and the encoded information of the adjacent blocks in the selected list is supplied to the encoded information storage memory 210.

  In the second embodiment, one adjacent block is selected from each block group as a representative reference adjacent block. However, a plurality of adjacent blocks may be selected. However, it is set within the range not exceeding the number of blocks in the block group.

(Example 3)
In the third embodiment, the reference adjacent block selection method is changed according to the prediction mode. This will be described using the arrangement of adjacent blocks in FIG. In the merge mode, the encoding information of one adjacent block is selected as a reference destination from the six adjacent blocks A0, A1, B0, B1, C0, and T. In the motion detection mode, among the six adjacent blocks A0, A1, B0, B1, C0, T, A0 and A1 are classified into a block group adjacent to the left side, B0, B1 and C0 are classified into a block group adjacent to the upper side, The block is selected as an adjacent block representing one from the block group. Encoding information of one adjacent block is selected as a reference destination from each representative block of the block group adjacent to the left side and the upper side and three adjacent blocks of T. That is, the merge mode is a mixed method in which the method of the first embodiment is applied, and the motion detection mode is a method of applying the method of the second embodiment.

  In the merge mode, since the inter-picture prediction is performed using the coding information of the reference adjacent block candidates as they are, more candidates are registered in the list than in the motion detection mode in order to leave room for selection. This is because the merge mode can transmit a lot of information (motion vector, reference index, reference list) with a small amount of code compared to the motion detection mode, rather than reducing the code amount of the merge index. This is because the coding efficiency improves when the selection range is widened by leaving the adjacent block candidates. On the other hand, in the motion detection mode, the encoding information including the differential motion vector is encoded instead of using the encoding information of the reference adjacent block candidates as they are, so that the candidates are registered in the list by performing scanning. . This is because the encoding efficiency is improved by reducing the code amount of the MVP index by narrowing down candidates.

  The candidate list created in the third embodiment is represented as shown in FIG. 41 for each prediction mode when all the reference neighboring blocks are valid. Reference candidate lists are respectively created by the merge detection unit 106 and the motion vector prediction unit 103, and the total number of candidate lists is also different, so the codewords are also different for the three or more items, but each is created as an optimal codeword. Improvement can be expected. In addition, although the codeword length becomes long, the codeword “11” (binary notation) in the motion detection mode is matched with “110” (binary notation) in the merge mode on the same line, and the type of codeword that appears is changed. It may be shared.

Example 4
The fourth embodiment is a method of selecting two blocks as neighboring blocks from neighboring blocks (hereinafter referred to as spatial neighboring blocks) on the same picture as the prediction block to be processed. This point is the same as in the second embodiment, but in the second embodiment, adjacent blocks in the space are defined by being divided into two block groups, and one adjacent block is selected for each block group. The fourth embodiment is different in that the entire adjacent blocks in the space are made one block group, and two representative adjacent blocks are selected from these blocks.

  As in the second embodiment, three adjacent blocks including two adjacent blocks representing adjacent blocks on the same picture as the prediction block to be processed and blocks adjacent to the same position of another picture having different times are included in the reference candidate list. Will be registered. Since the total number of adjacent blocks registered in the reference candidate list can be reduced compared to the first embodiment, the number of comparisons of encoded information in the same information detection unit 312 and the number of comparisons of motion vectors in the same information detection unit 322 Can be reduced. Furthermore, the codeword length assigned to the index of the reference candidate list is shortened, and the code amount of the index of the reference candidate list can be reduced. Furthermore, by changing the selection processing order of adjacent blocks according to the prediction mode, it is possible to select an adjacent block suitable for the prediction mode, and an improvement in coding efficiency can be expected.

  When the prediction mode is the merge mode, the merge detection unit 106 and the merge determination unit 206 are in the order of A1, B1, B0, A0, C0 (or B1, A1, A0, B0, C0) in FIG. Process with. In the motion detection mode, the motion vector prediction unit 103 and the motion vector calculation unit 204 perform processing in the order of B0, A0, A1, B1, C0 (or A0, B0, B1, A1, C0) in FIG. Do.

  In the merge mode, adjacent blocks A1 and B1 are preferentially selected and registered in the candidate list. This is because the adjacent blocks A1 and B1 are in contact with the prediction block to be processed, and the sides thereof are in contact with each other. Registering a candidate having a high and high possibility contributes to an improvement in encoding efficiency in the merge mode. On the other hand, in the motion detection mode, adjacent blocks A0 and B0 are preferentially selected and registered in the candidate list. The adjacent blocks A0 and B0 have the longest distance between candidates in the adjacent blocks, and it is highly possible that a motion vector having a different property can be registered, and the range of selection of a predicted motion vector can be expanded. This makes it difficult to generate a large amount of code when transmitting a differential motion vector, which contributes to improvement in coding efficiency in the motion detection mode.

  The configurations of the moving image encoding device 100 and the moving image decoding device 200 of the fourth embodiment are the same as those of the first and second embodiments. The operation of the fourth embodiment is the same as that of the second embodiment except for the operations of the encoded information deriving unit 310 in FIGS. 11 and 22 and the encoded information deriving unit 320 in FIGS. Hereinafter, only the operation of the fourth embodiment different from the second embodiment will be described in detail.

  First, the operation of the merge detection unit 106 in the video encoding device will be described. The merge detection unit 106 has the same configuration as that of FIG. 11 of the second embodiment, but the processing of the encoded information deriving unit 310 is different from that of the second embodiment, so the operation of the encoded information deriving unit 310 in the fourth embodiment will be described. . FIG. 42 is a flowchart illustrating the operation of the encoded information deriving unit 310 according to the fourth embodiment.

  First, the output destination of the inter prediction information detection unit 104 is switched to the merge detection unit 106 in accordance with the prediction mode controlled by the video encoding device 100 input to the switch 108, and the internal prediction information detection unit 104 has an internal destination. The encoding information of the prediction adjacent block to be processed, which is stored in the temporary memory 303, is input to the encoding information deriving unit 310.

  From the input adjacent blocks, it is determined whether or not the coding information of the adjacent block in the same picture as the prediction block to be processed is valid, and two adjacent blocks are derived (S105). FIG. 43 shows a detailed flow of determination processing for each adjacent block in the same picture as the prediction block to be processed.

  First, the variable F indicating the selected adjacent block is set to S0, and the counter k is set to 0 (S1600). S0 of the variable F indicating the selected adjacent block indicates the first candidate adjacent block selected first, and S1 indicates the second candidate adjacent block selected second.

  Next, a variable N indicating the adjacent block being processed is initialized to A1 (S1601). The variable N indicating the adjacent block being processed is set according to the processing order of A1, B1, B0, A0, C0, and is updated in this order when N is updated.

Next, a variable for storing coding information of adjacent blocks used in the subsequent determination is initialized (S1602). Here, the variables are a flag availableFlagF indicating whether the adjacent block is valid, a motion vector mvLXF, a reference picture number refIdxLXF, and a flag predFlagLXF indicating the validity of the prediction direction, and are initialized as follows.
availableFlagF = 0
mvLXF = (0,0)
refIdxLXF = 0
predFlagLXF = 0
Here, 0 or 1 indicating the prediction direction is set in the subscript X. The position of the adjacent block (hereinafter referred to as the adjacent block N) of the variable N indicating the adjacent block being processed and the encoding information are acquired (S1603).

  Based on the acquired position of the adjacent block N, it is determined whether or not the adjacent block N is valid (S1604). For example, when the prediction block to be processed is located at the left end of the picture, there is no adjacent block to the left of the prediction block to be processed, so there is no corresponding encoding information in the encoding information storage memory 114. It is determined to be invalid.

  When the adjacent block N is invalid (N in S1604), availableFlagF is set to “0” (S1613). When the adjacent block N is valid (Y in S1604), it is determined whether or not the prediction mode of the adjacent block N is the intra-frame coding (Intra) mode (S1605), and when the prediction mode of the adjacent block N is Intra mode ( Y of S1605), availableFlagF is set to “0” (S1613).

  When the prediction mode of the adjacent block N is not Intra mode (N in S1605), that is, in the Inter mode, the process proceeds to determination of whether F is S0 (S1606). If F is S0 (Y in S1606), availableFlagF (AvailableFlagS0) is set to 1 (S1608).

  If F is not S0 (N in S1606), it is determined whether the encoding information of the adjacent block already selected and the encoding information of the adjacent block N are the same (S1607). If the encoding information of the adjacent block already selected and the encoding information of the adjacent block N are the same (Y in S1607), availableFlagF is set to “0” (S1613).

  Specifically, when the maximum number of adjacent blocks in the space is set to 2, the encoding information first selected as the first candidate is retained and used for comparison determination when selecting the second candidate. By this process, the encoding information of the adjacent block of the second candidate does not overlap with the encoding information of the adjacent block of the first candidate that is representative, so that more candidates can be selected as reference destinations. In order to reduce the amount of processing, the processing in steps S1606 and S1607 can be omitted and the processing can proceed directly to step S1608.

  If the encoding information of the adjacent block selected as the first candidate is different from the encoding information of the adjacent block N (N in S1607), availableFlagF (availableFlagS1) is set to “1” (S1608). Subsequently, the encoding information of the adjacent block N is substituted and set in refIdxLXF, mvLXF, and predFlagLXF (S1609).

  Next, the counter k is updated by adding 1 (S1610), and it is determined whether the counter k is smaller than 2, which is the maximum number of candidates for the space (S1611). When the counter k is 2 which is the maximum number of space candidates (Y in S1611), a variable NumListCand indicating the number of adjacent block candidates in the selected space is set to k (S1616), and the process ends.

  When availableFlagF is set to “0” (S1613), it is determined whether or not the adjacent block N is the last of the adjacent blocks (S1614). If k is smaller than 2 which is the maximum number of space candidates (N in S1611), F is updated to S1 (S1612), and it is determined whether or not the adjacent block N is the last of the adjacent blocks ( S1614). The update of the variable N is set according to the processing order of the above-described A1, B1, B0, A0, C0, and here, it is determined whether or not N is C0.

  If the adjacent block N is the last block (Y in S1614), a variable NumListCand indicating the number of adjacent block candidates in the selected space is set to k (S1616), and the process ends. If the adjacent block N is not the last block (N in S1614), the variable N is updated according to the processing order of A1, B1, B0, A0, and C0 (S1615), and the processes after step S1602 are repeated. As described above, the respective pieces of encoding information are derived as two reference blocks representing adjacent blocks of a block group existing on the same picture as the prediction block to be processed.

  Returning to FIG. 42, following step S105, the reference picture number of the prediction block to be processed is determined when the coding information of the block adjacent to the same position of another picture with a different time is a candidate (S104). . The process in step S104 is the same as step S104 in the first and second embodiments. Next, it is determined whether or not the coding information of a block adjacent to the same position of another picture with different time is valid (S102). The processing in step S102 in the fourth embodiment is the same as that in step S102 in the first and second embodiments.

  The encoding information of the adjacent blocks obtained in this way is input to the reference candidate list creation unit 311. Also in the fourth embodiment, as in the second embodiment, a reference candidate list is provided in the reference candidate list storage memory 314 as a storage area for registering encoding information candidates of reference neighboring blocks. Priorities of encoding information candidates of reference adjacent blocks are given priority, and encoding information candidates of reference adjacent blocks are registered in the reference candidate list in descending order of priority. Thereby, the code amount of the index merge_idx in the reference candidate list is reduced.

  The reference candidate list creation unit 311 performs the same operation as in the first and second embodiments shown in the flowchart of FIG. However, the operation is performed by setting the value of the variable F of the fourth embodiment as the value of the variable N of FIG. The effective flag availableFlagN (where N is S0, S1, T) of the adjacent block S0 of the first candidate in space, the adjacent block S1 of the second candidate in space, and the block T adjacent to the same position of another picture with different time And the coding information of the adjacent block is registered in the reference candidate list candList. Here, the order of adjacent blocks in which the variable N is updated is the priority order for storing in the reference candidate list. In the fourth embodiment, the first candidate at the same time, the second candidate at the same time, and the same order at different times. The settings are made in the order of (S0, S1, T).

  As in the first and second embodiments, the created reference candidate list deletes the encoded information stored in the reference candidate list that is the same encoded information in the same information detection unit 312 and outputs it in the output unit 313. An index and encoding information in the created reference candidate list are output. The reference candidate list outputs a reference candidate list as a merge list and an index in the list as a merge index. In the moving image encoding apparatus 100, the video is output to the motion compensation prediction unit 105 and the prediction method determination unit 107.

  The merge determination unit 206 according to the fourth embodiment installed in the video decoding device 200 corresponding to the merge detection unit 106 according to the fourth embodiment installed in the above-described video encoding device 100 will be described. The merge determination unit 206 has the same configuration as that of FIG. 22 of the second embodiment, and the processing of the encoded information deriving unit 310 is different from that of the second embodiment. Other reference candidate list creation unit 311, identical information detection unit 312, reference candidate list storage memory 314 and selection unit 315 have the same functions as those in the second embodiment.

  The encoded information deriving unit 310 of the merge determining unit 206 has the same function as the encoded information deriving unit 310 of the merge detecting unit 106 in the fourth embodiment described above, and the reference candidate list creating unit of the merge determining unit 206 311 has the same function as the reference candidate list creation unit 311 of the merge detection unit 106 in the fourth embodiment, and the same information detection unit 312 of the merge determination unit 206 has the same function as that of the merge detection unit 106 in the fourth embodiment. Since the same function as that of the same information detection unit 312 is provided, the same reference candidate list as that of the merge detection unit 106 of the fourth embodiment is created by the processing up to the same information detection unit 312 of the merge determination unit 206.

  As in the second embodiment, the selection unit 315 selects a reference candidate list specified by an index that identifies the reference adjacent block decoded by the first encoded bit string decoding unit 202 from the generated reference candidate list. Select an adjacent block. The selected encoded information is supplied to the motion compensation prediction unit 207 and also supplied to the encoded information storage memory 210.

  Next, the motion vector predicting unit 103 according to the fourth embodiment installed in FIG. 2 illustrating the configuration of the moving image encoding device 100 will be described. The motion vector prediction unit 103 has the same configuration as that of FIG. 23 of the second embodiment, but the processing of the encoded information deriving unit 320 is different from that of the second embodiment, so the operation of the encoded information deriving unit 320 in the fourth embodiment will be described. To do. FIG. 44 is a flowchart illustrating the operation of the encoded information deriving unit 320 according to the fourth embodiment.

  First, the output destination of the inter prediction information detection unit 104 is switched to the motion vector prediction unit 103 according to the prediction mode controlled by the video encoding device 100 input to the switch 108. The encoding information of the processing target prediction block that is the reference adjacent block and stored in the temporary memory 303 inside the inter prediction information detection unit 104 is input to the encoding information deriving unit 320.

  From the input adjacent blocks, it is determined whether the coding information of the adjacent block in the same picture as the processing target prediction block is valid, and two adjacent blocks are derived (S803). FIG. 45 shows a detailed flow of determination processing for each adjacent block in the same picture as the prediction block to be processed. The process shown in FIG. 45 is performed in the case of L0 prediction, in the case of L1 prediction, and in the case of both predictions of L0 prediction and L1 prediction in accordance with the prediction mode of the block to be processed.

  First, the space candidate index F is set to S0, and the counter k is set to 0 (S1700). S0 of the spatial candidate index F indicates the first candidate adjacent block selected first, and S1 indicates the second candidate adjacent block selected second.

  Next, the variable N in the block group is initialized to A1 (S1701). The variable N indicating the adjacent block being processed is set according to the processing order of B0, A0, A1, B1, and C0, and is updated in this order when N is updated.

Next, a variable for storing encoding information of adjacent blocks used in the subsequent determination is initialized (S1702). Here, the variables are a flag availableFlagF indicating whether the adjacent block is valid, a motion vector mvLXF, a reference picture number refIdxLXF, and a flag predFlagLXF indicating the validity of the prediction direction, and are initialized as follows.
availableFlagLXF = 0
mvLXF = (0,0)
refIdxLXF = 0
predFlagLXF = 0
Here, 0 or 1 indicating the prediction direction is set in the subscript X. The position of the adjacent block of variable N (hereinafter referred to as adjacent block N) indicating the adjacent block being processed and the encoding information are acquired (S1703).

  Based on the acquired position of the adjacent block N, it is determined whether or not the adjacent block N is valid (S1704). For example, when the prediction block to be processed is located at the left end of the picture, there is no adjacent block to the left of the prediction block to be processed, so there is no corresponding encoding information in the encoding information storage memory 114. It is determined to be invalid.

  If the adjacent block N is invalid (N in S1704), the availableFlagLXF is set to “0” (S1715). If the adjacent block N is valid (Y in S1704), it is determined whether or not the prediction mode of the adjacent block N is the intra-frame coding (Intra) mode (S1705). If the prediction mode of the adjacent block N is Intra mode ( Y of S1705), availableFlagLXF is set to “0” (S1715).

  When the prediction mode of the adjacent block N is not the Intra mode (N in S1705), that is, in the Inter mode, the prediction target of the calculation target prediction block is compared with the encoding information of the adjacent block N from Condition 1. 4 is performed (S1709). When the condition 1 or 2 is satisfied, the process proceeds to a determination process of whether F is S0 (S1706). On the other hand, when the condition 3 or 4 is satisfied, the motion vector scaling processing of the adjacent block N is performed (S1710) similarly to S912 of FIG. 39 of the second embodiment, and the process proceeds to the determination processing of whether F is S0 ( S1706).

  When F is S0 (Y in S1706), availableFlagLXF (availableFlagLXS0) is set to 1 (S1708). If F is not S0 (N in S1706), whether or not the motion vector of the already selected adjacent block and the motion vector mvLXN in which the motion vector of the adjacent block N or the motion vector of the adjacent block N is scaled are the same. Determination is made (S1707).

  When the motion vector of the adjacent block already selected and the motion vector of the adjacent block N are the same (Y in S1707), availableFlagLXF is set to “0” (S1715). Specifically, when the maximum number of adjacent blocks in the space is 2, the motion vector that is first selected as the first candidate is stored and used for comparison determination when selecting the second candidate. By this processing, the encoded information of the adjacent block of the second candidate does not overlap with the motion vector of the adjacent block of the first candidate represented, so that the range of options as a reference destination is expanded. In order to reduce the processing amount, the processing in steps S1706 and S1707 may be omitted, and the process may directly proceed to step S1708 to set availableFlagLXF (availableFlagLXS1) to “1” (S1708).

  When the motion vector mvLXS0 of the adjacent block selected as the first candidate is different from the motion vector mvLXN of the adjacent block N (N in S1707), the availableFlagLXF (availableFlagLXS1) is set to “1” (S1708). Scaling of motion vector mvLX of the same reference list (when condition 1 is met) of adjacent block N or motion vector mvLY of a different reference list (when condition 2 is met) or the same reference list (when condition 3 is met) The motion vector mvLX converted by (1) or the motion vector mvLY converted by scaling of a different reference list (when condition 4 is met) is substituted into mvLXM (S1711).

  The counter k is updated by adding 1 (S1712), and it is determined whether the counter k is smaller than 2, which is the maximum number of candidates for the space (S1713). When the counter k is 2 which is the maximum number of space candidates (N in S1713), the variable NumListCand indicating the number of adjacent block candidates in the selected space is set to k (S1718), and the process ends.

  When availableFlagF is set to “0” (S1715), it is determined whether or not the adjacent block N is the last of the adjacent blocks (S1716). If k is smaller than 2 which is the maximum number of space candidates (Y in S1713), F is updated to S1 (S1714), and it is determined whether the adjacent block N is the last of the adjacent blocks ( S1716).

  The update of the variable N is set in accordance with the processing order of B0, A0, A1, B1, and C0 described above. Here, it is determined whether or not N is C0. If the adjacent block N is the last block (Y in S1716), the variable NumListCand indicating the number of adjacent block candidates in the selected space is set to k (S1718), and the process ends. If the adjacent block N is not the last block (N in S1716), the variable N is updated according to the processing order of B0, A0, A1, B1, and C0 (S1717), and the processing from step S1702 is repeated. As described above, the encoding information is derived as the reference block representing the adjacent block of the block group existing on the same picture as the prediction block to be processed.

  Returning to FIG. 44, next, it is determined whether or not the coding information of the block adjacent to the same position of another picture having a different time is valid (S801). Since the process of S801 of the fourth embodiment is the same as S801 of the first embodiment, a description thereof will be omitted.

  The encoding information of the adjacent blocks obtained in this way is input to the reference candidate list creation unit 321. The reference candidate list created by the reference candidate list creation unit 321 is one or both of the two prediction directions of L0 prediction and L1 prediction, depending on the prediction mode (L0 prediction, L1 prediction, both predictions). Each reference candidate list is created. In this specification, unless otherwise noted, only one reference candidate list in the motion detection mode will be described for convenience of explanation, and only one reference candidate list shown in the figure will be given.

  In the fourth embodiment, a reference candidate list is provided in the reference candidate list storage memory 324 as a storage area for registering encoding information candidates of reference neighboring blocks. Priorities of encoding information candidates of reference adjacent blocks are prioritized, and encoding information candidates of reference adjacent blocks are registered in the reference candidate list storage memory 324 in descending order of priority. Thereby, it is possible to reduce the code amount of the indexes mvp_idx_l0 and mvp_idx_l1 of the reference candidate list.

  The amount of codes is reduced by placing elements with higher priorities in front of the reference candidate list. The index indicating the priority order is assigned in ascending order from 0. The reference candidate list provided in the reference candidate list storage memory 324 has a list structure, and stores an index indicating the location inside the reference candidate list and encoding information candidates of reference adjacent blocks corresponding to the index as elements. An area is provided. This arrangement region is represented by candListLX. The number of the index starts from 0, and the encoding information candidates of the reference adjacent block are stored in the storage area of the reference candidate list candListLX.

  In the subsequent processing, the encoding information of the index i registered in the reference candidate list candListLX is represented by candListLX [i], and is distinguished from the reference candidate list candListLX by array notation. Note that the encoding information stored in the storage area of the reference candidate list is represented by the position name (A0, A1, B0, B1, C0, T) of the reference adjacent block unless otherwise specified.

  The reference candidate list creation unit 321 performs the same operation as in the first and second embodiments shown in the flowchart of FIG. However, the operation is performed by setting the value of the variable F of the fourth embodiment as the value of the variable N of FIG. The effective flag availableFlagLXN (where N is S0, S1, T) of the adjacent block S0 of the first candidate in space, the adjacent block S1 of the second candidate in space, and the block adjacent to the same position of another picture with different time A determination is made and the encoding information of the adjacent block is registered in the reference candidate list candListLX. Here, the order of adjacent blocks in which the variable N is updated is the priority order for storing in the reference candidate list. In the fourth embodiment, the first candidate at the same time, the second candidate at the same time, and the same order at different times. (S0, S1, T) is set.

  As in the first and second embodiments, the generated reference candidate list is obtained by deleting the encoded information stored in the reference candidate list that is the same encoded information in the same information detecting unit 322, and calculating the difference motion vector calculating unit. The motion vector of the encoded information in the reference candidate list created at 326 is set as a predicted motion vector. A difference motion vector is calculated from the motion vector detected by the motion vector detection unit 102 and the predicted motion vector, and the calculated difference motion vector is supplied to the output unit 323.

  Finally, the output unit 323 outputs the index and the difference motion vector in the created reference candidate list. The reference candidate list outputs the reference candidate list as an MVP list and the index in the list as an MVP index. In the moving image encoding apparatus 100, the image is output to the prediction method determination unit 107.

  The motion vector calculation unit 204 of Example 4 installed in the video decoding device 200 corresponding to the motion vector prediction unit 103 of Example 4 installed in the video encoding device 100 described above will be described. . The motion vector calculation unit 204 has the same configuration as that of FIG. 30 of the second embodiment, and the processing of the encoded information deriving unit 320 is different from that of the second embodiment. Other reference candidate list creation unit 321, identical information detection unit 322, reference candidate list storage memory 324, selection unit 325, and motion vector addition unit 327 have the same functions as those in the second embodiment.

  The encoded information deriving unit 320 has the same function as the encoded information deriving unit 320 of the motion vector predicting unit 103 in the second embodiment described above, and the reference candidate list creating unit 311 of the merge determining unit 206 is described above. The same function as the reference candidate list creation unit 311 of the merge detection unit 106 in the fourth embodiment is provided, and the same information detection unit 312 of the merge determination unit 206 is the same information detection unit of the merge detection unit 106 in the fourth embodiment described above. Since it has the same function as 312, the same reference candidate list as that of the motion vector prediction unit 103 of the fourth embodiment is created by the processing up to the same information detection unit 322 of the motion vector calculation unit 204.

  As in the second embodiment, the selection unit 325 selects a reference candidate list specified by an index that identifies a reference adjacent block decoded by the first encoded bit string decoding unit 202 from the generated reference candidate list. Select an adjacent block. The motion vector is output as the predicted motion vector from the coding information of the adjacent block of the selected reference candidate list, and the motion vector adding unit 327 decodes the predicted motion vector and the first coded bit string decoding unit 202. The motion vector is calculated by adding the vectors, supplied to the motion compensation prediction unit 207, and the encoded information of the adjacent blocks in the selected list is supplied to the encoded information storage memory 210.

  In the fourth embodiment, two adjacent blocks are selected as representative reference reference blocks. However, one or three or more adjacent blocks are selected as representatives for each of the merge mode and motion detection mode. I do not care. However, it is set within the range not exceeding the number of blocks in the block group. In the arrangement of FIG. 9 used for explaining the present invention, the maximum number of representatives is six. In the merge mode, setting the number of representatives to 6 is equivalent to the first embodiment. Adjacent blocks in two spaces are representative in motion detection mode, and adjacent blocks in three or more spaces are representative in merge mode. For example, adjacent blocks in more spaces are representative in merge mode than in motion detection mode. As a result, the following effects can be obtained.

  In other words, in the merge mode, inter-picture prediction is performed using the coding information of the reference adjacent block candidates as they are, so that more candidates are registered in the list than in the motion detection mode in order to leave room for selection. This is because the merge mode can transmit a large amount of information (motion vector, reference index, prediction direction) with a small amount of code compared to the motion detection mode, rather than reducing the code amount of the merge index. This is because the coding efficiency improves when the selection range is widened by leaving the adjacent block candidates.

  On the other hand, in the motion detection mode, encoding information including a differential motion vector is encoded instead of using the encoding information of the reference adjacent block candidates as they are, so that candidates are registered in the list more narrowly than in the merge mode. To do. This is because the encoding efficiency is improved by reducing the code amount of the MVP index by narrowing down candidates.

  In the merge mode, when it is assumed that adjacent blocks in three or more spaces are selected as representatives, in step S1607 of FIG. 43, the encoding information of all the already selected adjacent blocks and the encoding information of the adjacent block N are selected. Are the same. In the motion detection mode, when it is assumed that adjacent blocks in three or more spaces are selected as representatives, in step S1707 of FIG. 45, the motion vectors of all the already selected adjacent blocks and the motion vectors of adjacent blocks N or the adjacent blocks are selected. It is determined whether the motion vectors mvLXN to which the motion vectors of the block N are scaled are the same.

  In the merge mode, the method of the fourth embodiment can be used, and in the motion detection mode, the method of another embodiment can be used. Similarly, the method of the fourth embodiment can be used in the motion detection mode, and the method of the other embodiments can be used in the merge mode. In particular, when the method of the first embodiment is used in the merge mode and the method of the fourth embodiment is used in the motion detection mode, the same effect as that of the third embodiment can be obtained.

  In other words, in the merge mode, inter-picture prediction is performed using the coding information of the reference adjacent block candidates as they are, so that more candidates are registered in the list than in the motion detection mode in order to leave room for selection. This is because the merge mode can transmit a large amount of information (motion vector, reference index, prediction direction) with a small amount of code compared to the motion detection mode, rather than reducing the code amount of the merge index. This is because increasing the number of possible motion information candidates while leaving as many adjacent block candidates as possible contributes to an improvement in coding efficiency.

  On the other hand, in the motion detection mode, the encoding information including the differential motion vector is encoded instead of using the encoding information of the reference adjacent block candidates as they are, so that the candidates are registered in the list by performing scanning. . This is because the candidates in the motion detection mode are only used as prediction vectors, so even if many adjacent block candidates remain or even if motion vectors close to each other are registered, the code amount of the difference motion vector is reduced. It is not possible to increase the code amount of the MVP index. Therefore, reducing the code amount of the MVP index by narrowing down candidates to be registered in the list contributes to the improvement of the coding efficiency.

  In both the merge mode and motion detection mode, candidates to be registered in the list can be determined using only information on whether adjacent blocks can be used as reference destination candidates. Appropriate candidates can be registered in the list for each mode and motion detection mode.

  Further, in the fourth embodiment, only the blocks adjacent to the same position of different pictures having different times are registered separately at the end of the reference candidate list. Adjacent blocks on the same picture are likely to have the same encoding information, and blocks adjacent to the same position in different pictures with different times are likely to have different encoding information. It can be said that the registration effect is high. For example, even when all adjacent blocks on the same picture are intra, blocks that are adjacent to the same position in different pictures with different times are likely to have inter prediction coding information. However, a block adjacent to the same position of another picture with a different time may have the same or close encoding information as the processing target prediction block compared to an adjacent block on the same picture as the processing target prediction block. The priority is lowered because it is low.

(Example 5)
The merge detection unit 106 according to the fifth embodiment installed in FIG. 2 illustrating the configuration of the moving image encoding device 100 will be described. The merge detection unit 106 has the same configuration as that shown in FIG. 11 described in the first embodiment, but the reference candidate list creation unit 311 in the fifth embodiment creates a reference candidate list because the processing of the reference candidate list creation unit 311 is different from that in the first embodiment. The operation of the unit 311 will be described. The reference candidate list creation unit 311 provides a reference candidate list in the reference candidate list storage memory 314 as a storage area for registering encoding information candidates of reference adjacent blocks, and sets priorities to the encoding information candidates of reference adjacent blocks. In addition, encoding candidate information of reference neighboring blocks is registered in the reference candidate list in descending order of priority. Thereby, the code amount of the index merge_idx in the reference candidate list is reduced.

  The amount of codes is reduced by placing elements with higher priorities in front of the reference candidate list. The index indicating the priority order is assigned in ascending order from 0. Here, in the merge mode, the order (A1, B1, B0, A0, C0, T) is set. In the merge mode, a list is given with priority on adjacent blocks A1 and B1 in which the side of the prediction block to be processed is in contact with the side of the prediction block to be processed that is most likely to have the same encoding information. By registering in front of, the code amount of the merge index is reduced and the coding efficiency is improved.

  The reference candidate list provided in the reference candidate list storage memory 314 has a list structure, and stores an index indicating the location in the reference candidate list and encoding information candidates of reference adjacent blocks corresponding to the index as elements. An area is provided. This arrangement region is represented by candList. The index number starts from 0, and encoding information candidates of reference adjacent blocks are stored in the storage area of the reference candidate list candList. In the subsequent processing, the encoding information of the index i registered in the reference candidate list candList is represented by candList [i], and is distinguished from the reference candidate list candList by array notation. Note that the encoding information stored in the storage area of the reference candidate list is represented by the position name (A0, A1, B0, B1, C0, T) of the reference adjacent block unless otherwise specified.

  FIG. 46 is a flowchart illustrating the operation of the reference candidate list creation unit 311 in the fifth embodiment. In the fifth embodiment, before the reference candidate list is created, the adjacent blocks shown in FIG. 31A are classified into the left and upper block groups, and the numbers NA and NB of the reference adjacent blocks in each block group are set. To do. Here, the adjacent blocks of the left block group are A0, A1, the adjacent blocks of the upper block group are B0, B1, C0, and NA and NB are both set to 1. That is, it means selecting one adjacent block from each block group. Note that NA and NB do not exceed the total number of adjacent blocks in each block group.

  First, the variable N and the index k of the reference candidate list are initialized (S1200). In the variable N, an adjacent block with the highest priority corresponding to the prediction mode is initially set. It is A1 in the merge mode and A0 in the motion detection mode. k is set to zero. The index k indicates the priority order of the storage areas for the encoded information candidates set in the storage area of the reference candidate list.

  Next, the counters na and nb of the left and upper block groups are initialized to 0 (S1201). First, it is determined whether or not the adjacent block N is an adjacent block belonging to the left or upper block group (S1202). If it does not belong (N in S1202), the process proceeds to step S1206. If it belongs (Y in S1202), it is determined whether it belongs to the left or upper block group (S1203).

  In the case of the left side, the process proceeds to a comparison between the counter na and NA (S1204). If na is smaller than NA (Y in S1204), the process proceeds to step S1206. Otherwise (N in S1204), the process proceeds to step S1210. In the case of the upper side, the process proceeds to a comparison between the counter nb and the NB (S1205). If nb is smaller than the NB (Y in S1205), the process proceeds to Step S1206, otherwise (N in S1205).

  When na and nb are equal to or greater than NA and NB, this indicates that the number of adjacent blocks selected in the reference candidate list in each block group has been exceeded. In this case, whether or not to register in the reference candidate list is not determined. , Not registered.

  The validity flag availableFlagN of the adjacent block N is determined (S1206). If availableFlagN is 1 (Y in S1206), the encoding information of the adjacent block N is registered in the reference candidate list candListLX [k] (S1207), and k is updated (S1208). Further, if the adjacent block N belongs to the left or upper block group, the counter na or nb is updated (S1209). If availableFlagN is 0 (N in S1206), the process proceeds to step S1210 without registration in the reference candidate list.

  It is determined whether or not the adjacent block N is the last reference block (S1210). If it is the last block (Y in S1210), the index value is set to the total number of candidate lists NumListCand (S1211), and the process is terminated. If it is not the last block (Y in S1210), the variable N is updated (S1212), and the processes after step S1202 are repeated.

  Through the above processing, it is registered in the reference candidate list. Here, the maximum number registered in the reference candidate list is represented by NA + NB + 1. One candidate of different time is added to the initially set NA and NB. Here, since NA and NB are both 1, the maximum is 3. When only one reference adjacent block is valid, the maximum codeword length is 0. Therefore, no codeword is required, and the candidate of encoding information of an adjacent block determined to be valid is unique as a reference destination. To be determined.

  In the above processing, for example, when the valid flag availableFlagA1 of the adjacent block A1 is 0, the priorities are A1, B1, B0, A0, C0, T, so B1, A0, T are registered in the reference candidate list in order of priority. . When processing is performed under the same conditions in the second embodiment, A0, B1, and T are obtained. By preferentially registering adjacent blocks that are considered to be most likely to be the same as the prediction block to be processed and the encoding information while maintaining the original priority in the fifth embodiment, the reference candidate list is registered in the front. The code amount of the index is reduced and the encoding efficiency is improved.

  Further, in this case, three adjacent blocks of the adjacent block representing the block group adjacent to the left side and the upper side and the block adjacent to the same position of another picture having different times are registered in the reference candidate list. Since the total number of adjacent blocks can be reduced as compared with the first embodiment, the codeword length to be assigned is shortened, and the coding efficiency can be improved.

  The merge determination unit 206 of Example 5 installed in the video decoding device 200 corresponding to the merge detection unit 106 of Example 5 installed in the above-described video encoding device 100 will be described. The merge determination unit 206 has the same configuration as that of FIG. 22 described in the first embodiment, and the processing of the reference candidate list creation unit 311 is different from that of the first embodiment. The other encoded information deriving unit 310, the same information detecting unit 312, the reference candidate list storage memory 314, and the selecting unit 315 have the same functions as those in the first embodiment.

  In addition, since the reference candidate list creation unit 311 has the same function as the reference candidate list creation unit 311 of the merge detection unit 106 in the fifth embodiment described above, the processing up to the same information detection unit 312 of the merge determination unit 206 is performed. Thus, the same reference candidate list as that of the merge detection unit 106 of the fifth embodiment is created.

  The selection unit 315 that acquires the encoding information of the reference adjacent block in the merge mode from the created reference candidate list will be described. The selection unit 315 selects an adjacent block in the reference candidate list specified by an index for specifying the reference adjacent block, decoded by the first encoded bit string decoding unit 202, from the created reference candidate list. The selected encoded information is supplied to the motion compensation prediction unit 207 and also supplied to the encoded information storage memory 210.

  Next, the motion vector predicting unit 103 according to the fifth embodiment installed in FIG. 2 showing the configuration of the moving picture coding apparatus 100 of the present invention will be described. The motion vector prediction unit 103 has the same configuration as that of FIG. 23 described in the first embodiment, but the processing of the reference candidate list creation unit 321 is different from that of the first embodiment. Since the reference candidate list creation unit 321 has basically the same function as the reference candidate list creation unit 311 of the merge detection unit 106, it is determined whether the encoding information candidate of the reference adjacent block is valid. Since the operations to be registered in the list are the same, the description is omitted.

  The reference candidate list creation unit 321 is different from the reference candidate list creation unit 311 of the merge detection unit 106 in that encoding candidate information of reference neighboring blocks is registered in the reference candidate list provided in the reference candidate list storage memory 324. In case priority. In the motion detection mode, the reference candidate list creation unit 321 sets A0, B0, A1, B1, C0, and T in this order.

  The motion detection mode is a mode for transmitting a difference motion vector, and in order to widen the selection range of the predicted motion vector, the left and the left and the upper adjacent block candidate motion vector difference is increased. By registering the distance between the upper candidates at a distance, the coding amount of the difference motion vector is reduced and the coding efficiency is improved. The priority determination method in this motion detection mode has a different purpose from the merge mode in which adjacent blocks with the highest probability of the same encoded information are registered as priority candidates. As described above, by assigning priorities and registering encoding information candidates of reference adjacent blocks in the reference candidate list in descending order of priority, the code amount of the indices mvp_idx_l0 and mvp_idx_l1 in the reference candidate list is reduced. To do.

  The motion vector calculation unit 204 of Example 5 installed in the video decoding device 200 corresponding to the motion vector prediction unit 103 of Example 5 installed in the above-described video encoding device 100 is also described. Since the reference candidate list creation unit 321 of the motion vector calculation unit 204 has the same function as the reference candidate list creation unit 321 of the corresponding motion vector prediction unit 103, the same reference candidate list as the motion vector prediction unit 103 exists. Will be created.

  The selection unit 325 that acquires the coding information of the reference adjacent block in the motion detection mode from the created reference candidate list will be described. The selection unit 325 selects, from the created reference candidate list, an adjacent block in the list specified by an index that identifies the reference adjacent block decoded by the first encoded bit string decoding unit 202. A motion vector is output as a predicted motion vector from the encoded information of adjacent blocks in the selected list, and the motion vector adder 327 predicts the motion vector and the first encoded bit string decoder 202 decodes the motion vector. And the motion vector is calculated and supplied to the motion compensation prediction unit 207, and the encoded information of the adjacent block in the selected list is supplied to the encoded information storage memory 210.

  In the fifth embodiment, both NA and NB are set to 1. However, a number that does not exceed the total number of adjacent blocks in each block group may be set, and each setting may be changed in the prediction mode. .

(Example 6)
In the sixth embodiment, a block group of reference adjacent blocks different from that in the second embodiment is defined, and the method of selecting one of them as a representative adjacent block is changed. This will be described using the arrangement of adjacent blocks in FIG. The six adjacent blocks A0, A1, B0, B1, C0, and T arranged in FIG. 31A are different representative pictures in which A1 is a representative block adjacent to the left side, B1 is a representative block adjacent to the upper side, and time is different. The block T adjacent to the same position and the adjacent blocks A0, B0 and C0 positioned at the corner of the prediction block to be processed are defined as a block group adjacent to the corner. The block is selected as an adjacent block representing one from the block group. Encoding information of one adjacent block is selected as a reference destination from four adjacent blocks on the left side, the upper side, different times, and corners.

  Preferentially register adjacent blocks A1 and B1 where the sides of the prediction block to be processed and the sides of the prediction block to be processed that are most likely to have the same encoding information are in front of the reference candidate list In addition, this method is effective particularly in the merge mode because three adjacent blocks of corners having a relatively low possibility are grouped to reduce the code amount of the index and improve the encoding efficiency.

  The merge detection unit 106 according to the sixth embodiment installed in the video encoding device 100 and the reference candidate list creation unit 311 of the merge determination unit 206 according to the sixth embodiment installed in the corresponding video decoding device 200. Reference candidates of the motion vector predicting unit 103 according to the sixth embodiment installed in the video encoding device 100 and the motion vector calculating unit 204 according to the sixth embodiment installed in the corresponding video decoding device 200. In the list creation unit 321, adjacent blocks A 0, B 0 and C 0 positioned at the corner of the prediction block to be processed are defined as a block group adjacent to the corner. By setting priorities in the order of A1, B1, (B0, A0, C0) and T, and registering candidate encoding information of reference adjacent blocks in the reference candidate list in descending order of priority, the reference candidate list Reduce the code amount of the index. Here, the parentheses in the reference candidate list represent one adjacent block that is selected from the left or upper adjacent block group in the order of left to right in the parentheses.

  In the reference candidate list creation units 311 and 321, as described in the second embodiment, for example, as described in the second embodiment, the neighboring blocks in the block group are selected in the predetermined processing order as the selection of the neighboring block from the block group adjacent to the corner. to scan. It may be determined whether the adjacent block is valid and the condition that is not Intra mode is satisfied, and the first adjacent block that satisfies this condition may be selected. Here, the processing order is B0, A0, C0, but this order is not necessarily required. Further, the coding information of the adjacent blocks B0, A0, and C0 in the block group adjacent to the corner may be compared, and the neighboring blocks having the same coding information may be selected. That is, a majority decision is made in the block group, and the adjacent block having the same encoding information is selected as the adjacent block in the processing order B0, A0, C0. The reference candidate list created in the sixth embodiment is expressed as shown in FIG. 47 when all the reference neighboring blocks are valid.

(Example 7)
In the seventh embodiment, the number of reference neighboring blocks registered in the reference candidate list is limited. The number of reference adjacent blocks registered in the reference candidate list in the first and third embodiments (in the case of the merge mode) is 6 at the maximum in the case of the arrangement of the adjacent blocks in FIG. 9A or FIG. There are. If an adjacent block is selected from these, a code amount of 5 bits at the maximum is required for the index, so it cannot be said that the encoding efficiency is improved. Therefore, only the encoding information of the upper adjacent block in the reference candidate list created in the same manner as in the second, fourth, and the like is used.

  First, the merge detection unit 106 according to the seventh embodiment installed in FIG. 2 showing the configuration of the video encoding device 100 will be described. FIG. 48 shows a configuration of the merge detection unit 106 in which a reference candidate restriction unit 318 is added to the merge detection unit 106 of the video encoding device 100 of the first embodiment shown in FIG. 48 have the same functions as the units shown in FIG. 11, only the function of the reference candidate restriction unit 318 will be described here.

  The reference candidate restriction unit 318 receives the reference candidate list obtained by deleting the reference adjacent block having the same encoded information from the reference candidate list in the same information detection unit 312. The reference candidate restriction unit 318 leaves the adjacent blocks higher in the index than the set restriction number from the reference candidate list, and deletes the other blocks or determines that they are not selected. For example, in FIG. 20 showing an example of the reference candidate list in the case where the same encoded information does not exist in the first embodiment, if the number is limited to 3 in the case of the merge mode, the top 3 neighbors from index 0 to 2 Since the block is a reference candidate and the rest is not used, it is excluded from the reference candidates.

  As described above, adjacent blocks are registered as reference candidates by being limited to the reference candidate list, and the total number of reference adjacent blocks can be reduced as compared with the first embodiment. This shortens the encoding efficiency. In addition, since the restriction is performed by narrowing down to adjacent blocks having a high priority in the reference candidate list, it is possible to suppress a decrease in encoding efficiency. The limit number of the reference candidate list may be set in header information such as SPS (Sequence Parameter Set) and Slice Header, for example. Moreover, it may be set as an implicit condition on the encoding side and the decoding side, and the method is not limited as long as it is set so that no contradiction occurs between the encoding side and the decoding side. Moreover, you may set a limit number according to prediction mode, respectively. Furthermore, the limit number of the reference candidate list may be adaptively changed based on the encoding information of the encoded or decoded adjacent block that is referred to by the processing target prediction block.

  As described above, the method of limiting the reference candidates by providing the reference candidate restriction unit 318 immediately after the identical information detection unit 312 has been described. However, the reference candidate restriction unit 318 is not provided in the merge detection unit 106. Even if the same information detection unit 312 is provided with the function of the reference candidate restriction unit 318, the same effect can be obtained. In this case, the same information detection unit 312 compares the encoded information candidates stored in the reference candidate list, and if the same encoded information candidate exists, the code having the smallest reference candidate list index is stored. Are deleted, except for encoding information candidates, and the encoding information candidates with the smallest reference candidate list index are re-registered in the reference candidate list, and the process ends when the number of registrations reaches the limit number To do.

  FIG. 49 is a flowchart showing the operation of the same information detection unit 312 in the seventh embodiment. The same information detection unit 312 that performs the operation shown in FIG. Is added. The processing steps from Step S710 to Step S712 are newly added to the flowchart of FIG. 21, and the added processing will be described.

  After setting the variables n and m representing the index of the reference candidate list, a counter s representing the number of reference candidate lists recorded in the deletion list is set to 0 (S710). Next, in step S702, it is determined whether m has already been recorded in the deletion list. If m has not yet been recorded in the deletion list (N in S702), m is recorded in the deletion list and is stored in the counter s. Update by adding 1 (S711).

  Next, it is determined whether or not s is equal to the threshold value TH (S712). Here, the threshold value TH represents a number obtained by subtracting the limit number from the total number of candidate lists NumListCand. That is, it represents the maximum number of candidates recorded in the deletion list while leaving a limited number of candidates in the reference candidate list. When s becomes equal to TH (Y in S712), a limited number of candidates remain in the reference candidate list, so that the same information determination process is terminated, and the process proceeds to step S708 to be recorded in the deletion list. The encoded information in the storage area of the list corresponding to the index is deleted, and the code word is updated in the order of candidates with the smallest index based on the index 0, and the process ends. The total number NumListCand of the reference candidate list is updated to the limit number.

  Further, the reference candidate restriction unit 318 may be provided immediately before the same information detection unit 312. In this case, the number of candidates is narrowed down by limiting the reference candidates in the reference candidate list created by the reference candidate list creation unit 311, so the maximum number of times the same information detection unit 312 detects the same information is reduced. There is an effect that the processing amount is reduced.

  FIG. 50 shows a configuration of the merge detection unit 106 in which the reference candidate restriction unit 318 of the seventh embodiment is added based on the merge detection unit 106 of the video encoding device 100 of the first embodiment shown in FIG. Each part shown in FIG. 50 has the same function as each part shown in FIG. The configuration of FIG. 48 described above and the position of the reference candidate restriction unit 318 are different, and the operation of the reference candidate restriction unit 318 is the same. The reference candidate restriction unit 318 leaves the adjacent blocks higher in the index for the set number of restrictions from the reference candidate list created by the reference candidate list creation unit 311 and deletes other blocks or excludes the selection. judge.

  As described above, the method of restricting reference candidates by providing the reference candidate restriction unit 318 immediately before the same information detection unit 312 has been described, but the reference candidate restriction unit 318 is not provided in the merge detection unit 106. Even if the function of the reference candidate restriction unit 318 is provided in the reference candidate list creation unit 311, the same effect can be obtained. In this case, the reference candidate list creating unit 311 registers the coding information of the reference neighboring block derived by the coding information deriving unit 310 in the reference candidate list and the reference neighboring registered in the reference candidate list. The number of blocks is counted, and the process ends when the number reaches the limit.

  FIG. 51 is a flowchart showing the operation of the reference candidate list creation unit 311 according to the seventh embodiment. The reference candidate list creation unit 311 that performs the operation shown in FIG. The function to perform is added. A new processing step of step S607 is added to the flowchart of FIG. 19, and the added processing will be described.

  First, the variable N and the index k of the reference candidate list are initialized (S600). The index k indicates the priority order of the storage areas for the encoded information candidates set in the storage area of the reference candidate list.

  Next, the valid flag availableFlagN of the reference adjacent block N is determined (S601). If availableFlagN is 1 (Y in S601), the encoding information of the adjacent block N is registered in the reference candidate list (S602), k Is updated (S603).

  Next, it is determined whether or not k has reached the limit number (S607). If k is less than the limit number (N in S607), the encoding information for the limit number has not yet been registered in the reference candidate list, and the process proceeds to step S604. If k reaches the limit number (Y in S607), the encoding information for the limit number is registered in the reference candidate list, so the index value is set to the total number of candidate lists NumListCand (S605), and the process ends. To do.

  In addition, although the example in which the reference candidate restriction unit 318 is added to the merge detection unit 106 of the first embodiment has been described, the reference candidate restriction unit 318 can be added also in other embodiments. It is also possible to add a reference candidate restriction unit 318 to the merge determination unit 206 of the seventh embodiment installed in the video decoding device 200 corresponding to the merge detection unit 106. Even if the reference candidate restriction unit 318 is not provided and the function of the reference candidate restriction unit 318 is provided in the reference candidate list creation unit 311 or the same information detection unit 312 of the merge determination unit 206, the same effect can be obtained. However, the position where the reference candidate restriction unit 318 is installed on the encoding side and the decoding side or the processing unit having the function of the reference candidate restriction unit 318 needs to be the same.

  Next, the motion vector predicting unit 103 according to the seventh embodiment installed in FIG. 2 illustrating the configuration of the moving image encoding device 100 will be described. FIG. 52 shows a configuration of the motion vector predicting unit 103 in which the reference candidate restricting unit 328 of the seventh embodiment is added based on the motion vector predicting unit 103 of the moving picture coding apparatus of the first embodiment shown in FIG. The units shown in FIG. 52 basically have the same functions as the units shown in FIG. 23, but the processing of the encoded information deriving unit 320 and the reference candidate list creating unit 321 is different from that of the first embodiment. The functions of the encoding information deriving unit 320 and the reference candidate list creating unit 321 and the newly added reference candidate restricting unit 328 according to the seventh embodiment will be described.

  FIG. 53 is a flowchart showing the operation of the encoded information deriving unit 320 in the seventh embodiment. In the encoded information deriving unit 320 according to the first embodiment, in accordance with the prediction mode of the prediction block to be processed, in the case of unidirectional prediction and L0 prediction, only the encoded information registered in the reference list L0 is unidirectionally predicted. In the case of L1 prediction, only the encoded information registered in the reference list L1 is determined. In the case of bi-prediction, the determination of the respective encoded information registered in the reference lists L0 and L1 is performed for each list of L0 and L1. Went independently.

  In the seventh embodiment, first, the validity of adjacent blocks is determined regardless of the list, and thereafter, encoding information is detected and acquired according to the prediction mode of the prediction block to be processed. With this processing, it is not necessary to determine the validity of adjacent blocks for each list, and the number of processing steps can be reduced.

  First, the variable N is initialized (S900). In the variable N, adjacent blocks B0, A0, A1, B1, and C0 on the same picture shown in FIG. Here, N = B0 is set at the time of initialization, and the variable N is updated in the order of A0, A1, B1, C0.

  The motion detection mode is a mode for transmitting a difference motion vector, and in order to widen the selection range of the predicted motion vector, the left and the left and the upper adjacent block candidate motion vector difference is increased. By registering the distance between the upper candidates at a distance, the coding amount of the difference motion vector is reduced and the coding efficiency is improved. The priority determination method in this motion detection mode has a different purpose from the merge mode in which adjacent blocks with the highest probability of the same encoded information are registered as priority candidates.

  Next, the position of the adjacent block of variable N (hereinafter referred to as adjacent block N) and encoding information are acquired (S902). Based on the acquired position of the adjacent block N, it is determined whether or not the adjacent block N is valid (S903). For example, when the prediction block to be processed is located at the left end of the picture, there is no adjacent block to the left of the prediction block to be processed, so there is no corresponding encoding information in the encoding information storage memory 114. It is determined to be invalid.

  When the adjacent block N is invalid (N in S903), the available FlagN is set to “0” (S908). When the adjacent block N is valid (Y in S903), it is determined whether or not the prediction mode of the adjacent block N is an intra-frame coding (Intra) mode (S904). When the prediction mode of the adjacent block N is not the Intra mode (Y in S904), that is, in the case of the Inter mode, the available FlagN is set to “1” (S906). Subsequently, a motion vector is calculated (S916). After calculating the motion vector, the encoding information of the adjacent block N is set by substituting it into refIdxLXN, mvLXN, and predFlagLXN (S907).

  When the determination process for the adjacent block N is completed as described above, it is determined whether or not the variable N is the last of the adjacent blocks (S909). Since the variable N is updated in the order of B0, A0, A1, B1, and C0, it is determined here whether N is C0 or not. If N is C0 (Y in S909), all adjacent blocks have been determined, and the process ends. If N is not C0 (N in S909), N is updated (S910). N is updated in the order of the adjacent blocks described above, and the processes in and after step S902 are repeated for the adjacent block N.

Here, the motion vector calculation process will be described with reference to the flowchart of FIG. First, a variable for storing the encoding information of the adjacent block N for L0 prediction is initialized. Here, the variables are the motion vector mvL0N of the adjacent block N, the reference picture number refIdxL0N, and the flag predFlagL0N indicating the validity of the reference list, and are initialized as follows (S920).
mvL0N = (0,0)
refIdxL0N = 0
predFlagL0N = 0)

  Next, it is determined whether the Inter mode of the prediction block to be processed is unidirectional L0 prediction or Bi-pred prediction (bi-prediction) (S921). Since Bi-pred prediction uses two reference lists of L0 prediction and L1 prediction, in the case of unidirectional L0 prediction or Bi-pred prediction (Y in S921), the process proceeds to calculation of the motion vector mvL0N of the adjacent block N ( S922).

  When the reference picture number and reference list of the prediction block to be processed and the adjacent block N are the same, the L0 prediction motion vector of the adjacent block N is substituted into mvL0N and set. When the prediction block to be processed and the reference picture number and reference list of the adjacent block N are not the same, the scaling processing of the motion vector of the adjacent block N is performed. Since the scaling process has been described in the first embodiment, it is omitted here. By performing the scaling process, it is possible to derive a closer motion vector for the prediction block to be processed and improve accuracy. As described above, the scaled motion vector is set to mvL0N, and the L0 prediction encoding information of the adjacent block N is set to refIdxL0N and predFlagL0N.

When neither the unidirectional L0 prediction nor the Bi-pred prediction is performed (N in S921), the process proceeds to the motion vector determination of the L1 prediction. Subsequently, a variable for storing the encoding information of the adjacent block N of the L1 prediction is initialized. Here, the variables are the motion vector mvL1N of the adjacent block N, the reference picture number refIdxL1N, and the flag predFlagL1N indicating the validity of the reference list, and are initialized as follows (S923).
mvL1N = (0,0)
refIdxL1N = 0
predFlagL1N = 0

  Next, it is determined whether the Inter mode of the prediction block to be processed is unidirectional L1 prediction or Bi-pred prediction (bi-prediction) (S924). Since Bi-pred prediction uses two reference lists of L0 prediction and L1 prediction, in the case of unidirectional L1 prediction or Bi-pred prediction (Y in S924), the process proceeds to calculation of a motion vector mvL1N of an adjacent block N ( S925).

  When the reference picture number and the reference list of the prediction block to be processed and the adjacent block N are the same, the motion vector of the L1 prediction of the adjacent block N is substituted into mvL0N and set. When the prediction block to be processed and the reference picture number and reference list of the adjacent block N are not the same, the scaling processing of the motion vector of the adjacent block N is performed. Since the scaling process has been described in the first embodiment, it is omitted here. As described above, the scaled motion vector is set in mvL1N, and the L1 prediction coding information of the adjacent block N is substituted into refIdxL1N and predFlagL1N. If neither the unidirectional L1 prediction nor the Bi-pred prediction is made (N in S924), the process is terminated.

  As described above, the encoding information is derived using the neighboring neighboring blocks existing on the same picture as the prediction block to be processed as the reference block. Since the derivation of the coding information of the block adjacent to the same position of another picture having a different time is the same as in the first embodiment, the description is omitted here.

  The encoding information of the adjacent blocks obtained in this way is input to the reference candidate list creation unit 321. As reference candidate lists created by the reference candidate list creation unit 321, reference candidate lists are created respectively for the two reference lists for L0 and L1 prediction according to the Inter mode of the prediction block to be processed. In the first embodiment, the reference candidate list is created for each reference list. However, in the seventh embodiment, the flag indicating the validity of the adjacent block used for the subsequent determination is represented by one flag availableFlagN regardless of the reference list. It is possible to create a reference candidate list for the L1 bi-prediction at a time, thereby reducing the number of processing steps.

  The operation of the reference candidate list creation unit 321 will be described with reference to the flowchart of FIG. First, the variable N and the index k of the reference candidate list are initialized (S600). In the variable N, the adjacent block B0 shown in FIG. 9A is initially set, and k is set to 0. The variable N represents the priority order stored in the reference candidate list, and is performed in the order of adjacent blocks B0, A0, A1, B1, and C0. The index k represents the number of the storage area of the encoding information candidate set in the storage area of the reference candidate list, and corresponds to the priority order.

  Next, the valid flag availableFlagN of the reference adjacent block N is determined (S601). When availableFlagN is 1 (Y in S601), the encoding information of the adjacent block N is registered simultaneously in the two reference candidate lists candListLX [k] for L0 and L1 prediction (S602), and k is updated (S603). Here, the subscript X represents 0 or 1. If availableFlagN is 0 (N in S601), it is not registered in the reference candidate list, and proceeds to the next. It is determined whether or not the adjacent block N is the last reference block (S604). If it is the last block (Y in S604), the index value is set to the total number of candidate lists NumListCand (S605), and the process ends. If it is not the last block (N in S604), the variable N is updated (S606), and the processes after step S601 are repeated.

  The reference candidate restriction unit 328 receives the reference candidate list obtained by deleting the reference adjacent block having the same motion vector from the reference candidate list by the same information detection unit 322. The reference candidate restriction unit 328 leaves the adjacent blocks that are higher in the index for the set restriction number from the reference candidate list, and deletes the other blocks or determines that they are not selected. For example, in FIG. 28 showing an example of the reference candidate list in the case where the same motion vector does not exist in the first embodiment, if the number is limited to 3 in the motion detection mode, the upper 3 adjacent blocks from index 0 to 2 Is not used as a reference candidate because the rest is not used.

  As described above, adjacent blocks are registered as reference candidates by being limited to the reference candidate list, and the total number of reference adjacent blocks can be reduced as compared with the first embodiment. This shortens the encoding efficiency. In addition, since the restriction is limited to adjacent blocks with high priority in the reference candidate list, it is possible to suppress a reduction in encoding efficiency. The limit number of the reference candidate list may be set in header information such as SPS (Sequence Parameter Set) and Slice Header, for example.

  Moreover, it may be set as an implicit condition on the encoding side and the decoding side, and the method is not limited as long as it is set so that no contradiction occurs between the encoding side and the decoding side. Moreover, you may set a limit number according to prediction mode, respectively. Furthermore, the limit number of the reference candidate list may be adaptively changed based on the encoding information of the encoded or decoded adjacent block that is referred to by the processing target prediction block.

  As described above, the method of limiting the reference candidates by providing the reference candidate restriction unit 328 immediately after the same information detection unit 322 has been described. However, the reference candidate restriction unit 328 is not provided in the motion vector prediction unit 103. In addition, even if the function of the reference candidate restriction unit 328 is provided in the same information detection unit 322, the same effect can be obtained. In this case, the same information detection unit 322 compares the encoded information candidates stored in the reference candidate list, and if there is a candidate having the same motion vector, the code having the index of the smallest reference candidate list All of them except for the candidates for encoding information are deleted, and the candidates for encoding information having the smallest index of the reference candidate list are re-registered in the reference candidate list, and the processing ends when the registration number reaches the limit number .

  FIG. 56 is a flowchart showing the operation of the same information detection unit 322 according to the seventh embodiment. The function of limiting the reference candidates of the seventh embodiment to the same information detection unit 322 that performs the operation shown in FIG. 29 of the first embodiment. Is added. The processing steps from step S1010 to step S1012 are newly added to the flowchart of FIG. 29, and the added processing will be described.

  After setting the variables n and m representing the index of the reference candidate list, a counter s representing the number of reference candidate lists recorded in the deletion list is set to 0 (S1010). Next, in step S1002, it is determined whether m is already recorded in the deletion list. If m is not yet recorded in the deletion list (N in S1002), m is recorded in the deletion list (S1003). The counter s is incremented by 1 and updated (S1011).

  Next, it is determined whether s is equal to the threshold value TH (S1012). Here, the threshold value TH represents a number obtained by subtracting the limit number from the total number of candidate lists NumListCand. That is, it represents the maximum number of candidates recorded in the deletion list while leaving a limited number of candidates in the reference candidate list. If s becomes equal to TH (Y in S1012), a limited number of candidates remain in the reference candidate list, so the same information determination process is terminated, and the process proceeds to step S1008 to be recorded in the deletion list. The encoded information in the storage area of the list corresponding to the index is deleted, and the code word is updated in the order of candidates with the smallest index based on the index 0, and the process ends. The total number NumListCand of the reference candidate list is updated to the limit number.

  Further, the reference candidate restriction unit 328 may be provided immediately before the same information detection unit 322. In this case, since the number of candidates is narrowed down by limiting the reference candidates in the reference candidate list created by the reference candidate list creation unit 321, the maximum number of times the same information is detected by the same information detection unit 322 is reduced. There is an effect that the processing amount is reduced.

  FIG. 57 shows a configuration of the motion vector prediction unit 103 in which the reference candidate restriction unit 328 of the seventh embodiment is added based on the motion vector prediction unit 103 of the moving picture coding apparatus of the first embodiment shown in FIG. Each part shown in FIG. 57 has the same function as each part shown in FIG. The configuration of FIG. 52 described above and the position of the reference candidate restriction unit 328 are different, and the operation of the reference candidate restriction unit 328 is the same. The reference candidate restriction unit 328 leaves the adjacent blocks that are higher in the index for the set restriction number from the reference candidate list created by the reference candidate list creation unit 321 and deletes other blocks or excludes them from selection. judge.

  As described above, the method of restricting reference candidates by providing the reference candidate restriction unit 328 immediately before the same information detection unit 322 has been described. However, the reference candidate restriction unit 328 is not provided in the motion vector prediction unit 103. Moreover, even if the reference candidate list creation unit 321 is provided with the function of the reference candidate restriction unit 328, the same effect can be obtained. In this case, the reference candidate list creation unit 321 registers the coding information of the reference neighboring block derived by the coding information deriving unit 320 in the reference candidate list and also the reference neighboring block registered in the reference candidate list. The number is counted, and the process is terminated when the number reaches the limit number. The reference candidate list creation unit 321 according to the seventh embodiment, except that the reference candidate list creation unit 321 can simultaneously create a reference candidate list for L0 and L1 prediction according to the prediction mode of the prediction block to be processed. Since the same operation as that of the flowchart of FIG. 51 showing the operation of FIG.

(Example 8)
In the eighth embodiment, a block group of reference adjacent blocks different from those in the second and fifth embodiments is defined, and K representative adjacent blocks are selected from the block group. This will be described using the arrangement of adjacent blocks in FIG. Similar to the fourth embodiment, from the six adjacent blocks A0, A1, B0, B1, C0, T arranged in FIG. 31A, adjacent blocks A0, A1, which are on the same picture as the prediction block to be processed. B0, B1, and C0 are set as one block group, and K representative adjacent blocks are selected from the block group.

  Encoding information of K + 1 representative blocks on the same selected picture and K + 1 adjacent blocks of block T adjacent to the same position of another picture with different time is registered as a reference candidate in the reference candidate list. . In the fourth embodiment, the coded information deriving unit 310 and the coded information deriving unit 320 select K (two) representative neighboring blocks, whereas in the eighth embodiment, as in the first embodiment, the code The encoding information deriving unit 310 and the encoded information deriving unit 320 are different in that they select all adjacent blocks and select K (two) representative adjacent blocks in the processing after the reference candidate list creation. .

  The eighth embodiment is equivalent to the case where the seventh embodiment that limits the number of adjacent blocks to be selected as reference candidates is applied to a block group on the same picture as the prediction block to be processed. The only difference is that only blocks adjacent to the same position in the picture are treated separately and registered at the end of the reference candidate list. Adjacent blocks on the same picture are likely to have the same encoding information, and blocks adjacent to the same position in different pictures with different times are likely to have different encoding information. It can be said that the registration effect is high. For example, even when all adjacent blocks on the same picture are intra, blocks that are adjacent to the same position in different pictures with different times are likely to have inter prediction coding information. However, a block adjacent to the same position of another picture with a different time may have the same or close encoding information as the processing target prediction block compared to an adjacent block on the same picture as the processing target prediction block. The priority is lowered because it is low.

  Therefore, the priority is set lower than that of the adjacent block selected from the block group on the same picture, and it is registered at a lower position in the reference candidate list. In the following description, the eighth embodiment is assumed to have the same configuration as that of the moving image encoding device 100 and the moving image decoding device 200 described in the seventh embodiment. However, the reference candidate restriction units 318 and 328 of the seventh embodiment are not installed, and the function of the reference candidate restriction unit is provided in the reference candidate list creation units 311 and 321 or the same information detection units 312 and 322, and the processing target It is applied to a block group of adjacent blocks on the same picture as the prediction block.

  Of the merge detection unit 106 installed in the video encoding device 100, the reference candidate list creation unit 311 of the merge determination unit 206 installed in the corresponding video decoding device 200, and the video encoding device The same picture as the prediction block to be processed in the motion vector prediction unit 103 installed in 100 and the reference candidate list creation unit 321 of the motion vector calculation unit 204 installed in the corresponding video decoding device 200 The adjacent blocks A0, A1, B0, B1, and C0 above are defined as a block group on the same picture. By setting priorities in the order of adjacent blocks representing block groups on the same picture, T, and by registering encoding information candidates of reference adjacent blocks in the reference candidate list in descending order of priority, Reduce the code amount of the index.

  First, the merge detection unit 106 according to the eighth embodiment installed in FIG. 2 illustrating the configuration of the moving image encoding device 100 will be described. The merge detection unit 106 has the same configuration as that of FIG. 11 described in the first embodiment, but the processing of the reference candidate list creation unit 311 is different from that of the first embodiment. The operation of the reference candidate list creation unit 311 of the merge detection unit 106 in the eighth embodiment will be described.

  FIG. 58 is a flowchart showing the operation of the reference candidate list creation unit 311 in the eighth embodiment. In the eighth embodiment, before the reference candidate list is created, the number K to be registered in the reference candidate list of the reference adjacent block of the block group on the same picture is set. Here, K is set to 2. Note that K does not exceed the total number of adjacent blocks in the block group on the same picture.

  First, the variable N and the index k of the reference candidate list are initialized (S1300). In the variable N, an adjacent block with the highest priority corresponding to the prediction mode is initially set. Since the priorities in the merge mode are A1, B1, B0, A0, C0, A1 is set, and N is updated in this order. k is set to zero. The index k indicates the priority order of the storage areas for the encoded information candidates set in the storage area of the reference candidate list.

  First, the validity flag availableFlagN of the adjacent block N is determined (S1301). If availableFlagN is 0 (N in S1301), the process proceeds to step S1305 without registering in the reference candidate list. When availableFlagN is 1 (Y in S1301), the encoding information of the adjacent block N is registered in the reference candidate list (S1302), and k is updated (S1303).

  After updating k, the process proceeds to comparison between k and K (S1304). If k is less than K (Y in S1304), that is, if the number of adjacent blocks registered in the reference candidate list has not reached the set number K, the process proceeds to step S1305. If k is greater than or equal to K (N in S1304), since the number of adjacent blocks registered in the reference candidate list has reached the set number K, the determination as to whether or not to register in the reference candidate list is skipped, and the process proceeds to step S1307. move on.

  Next, it is determined whether or not the adjacent block N is the last reference block (S1305). If the adjacent block N is not the last block (N in S1305), the variable N is updated (S1306), and the processes after step S1301 are repeated. If the adjacent block N is the last block (Y in S1305), the process proceeds to step S1307.

  After the registration determination of the block group on the same picture is completed, the valid flag availableFlagT of the block T adjacent to the same position of another picture having a different time is determined (S1307). When availableFlagT is 1 (Y in S1307), the encoding information of the adjacent block T is registered in the reference candidate list (S1308), and k is updated (S1309). If availableFlagT is 0 (N in S1307), the process proceeds to step S1310 without being registered in the reference candidate list. After the number registered in the reference candidate list is set as the total number of candidate lists NumListCand (S1310), the process ends.

  Through the above processing, it is registered in the reference candidate list. Here, the maximum number registered in the reference candidate list is represented by K + 1. When only one reference adjacent block is valid, the maximum codeword length is 0. Therefore, a codeword is not required, and a candidate of encoding information of an adjacent block determined to be only one is unique as a reference destination. To be determined.

  The merge determination unit 206 according to the eighth embodiment installed in the video decoding device 200 corresponding to the merge detection unit 106 installed in the above-described video encoding device 100 will be described. The merge determination unit 206 has the same configuration as that of FIG. 22 described in the first embodiment, and the processing of the reference candidate list creation unit 311 is different from that of the first embodiment. The other encoded information deriving unit 310, the same information detecting unit 312, the reference candidate list storage memory 314, and the selecting unit 315 have the same functions as those in the first embodiment.

  Further, since the reference candidate list creation unit 311 has the same function as the reference candidate list creation unit 311 of the merge detection unit 106 in the above-described eighth embodiment, the processing up to the same information detection unit 312 of the merge determination unit 206 is performed. Thus, the same reference candidate list as the merge detection unit 106 of the eighth embodiment is created. The selection unit 315 that acquires the encoding information of the reference adjacent block in the merge mode from the created reference candidate list will be described. The selection unit 315 selects an adjacent block in the reference candidate list specified by an index for specifying the reference adjacent block decoded by the first encoded bit string decoding unit 202 from the created reference candidate list. The selected encoded information is supplied to the motion compensation prediction unit 207 and also supplied to the encoded information storage memory 210.

  Next, the motion vector predicting unit 103 according to the eighth embodiment installed in FIG. 2 illustrating the configuration of the moving image encoding device 100 will be described. The motion vector prediction unit 103 has the same configuration as that of FIG. 23 described in the first embodiment, but the processing of the reference candidate list creation unit 321 is different from that of the first embodiment. The operation of the reference candidate list creation unit 321 in the eighth embodiment will be described.

  Similar to the reference candidate list creation unit 311 of the merge detection unit 106 described above, the reference candidate list creation unit 321 converts adjacent blocks A0, A1, B0, B1, and C0 on the same picture as the processing target prediction block into one block. As a group, K representative neighboring blocks are selected from the block group. Compared to the reference candidate list creation unit 311 of the merge detection unit 106, the reference candidate list for each of the L0 and L1 predictions is created at the same time. However, since there is no great difference in the flow of processing, the explanation is omitted here. To do. The priorities in the motion detection mode are B0, A0, A1, B1, and C0.

  The motion vector calculation unit 204 of the eighth embodiment installed in the video decoding device 200 corresponding to the motion vector prediction unit 103 installed in the video encoding device 100 is also a motion vector calculation unit. The reference candidate list creation unit 321 of 204 has the same function as the reference candidate list creation unit 321 of the corresponding motion vector prediction unit 103, and the same reference candidate list as the motion vector prediction unit 103 is created. Therefore, the explanation here is omitted.

  With the above processing, the candidate list created in the eighth embodiment is represented as shown in FIG. 59 for each prediction mode when all the reference adjacent blocks are valid. This is an example in which the number of reference adjacent blocks in the block group on the same picture is 2 in both the merge mode and the motion detection mode, and effective adjacent blocks with high priority in each prediction mode are registered in the reference candidate list. .

  In the eighth embodiment, the priority order is A1, B1, B0, A0, C0 in the merge mode, and B0, A0, A1, B1, C0 in the motion detection mode. There is no. However, in the merge mode, priority is given to adjacent blocks A1 and B1 in which the sides of the prediction block to be processed and the sides of the prediction block to be processed that are most likely to have the same encoding information are in contact. By registering in front of the reference candidate list, it is possible to reduce the code amount of the merge index and improve the encoding efficiency.

  The motion detection mode is a mode for transmitting a difference motion vector, and in order to widen the selection range of the predicted motion vector, the left and the left and the upper adjacent block candidate motion vector difference is increased. By registering the distance between the upper candidates apart, it is possible to reduce the coding amount of the differential motion vector and improve the coding efficiency. The priority determination method in this motion detection mode has a different purpose from the merge mode in which adjacent blocks with the highest probability of the same encoded information are registered as priority candidates. As described in the third embodiment, the combination is changed according to the prediction mode, such as using the eighth embodiment in the merge mode and using the second or other embodiments in the motion detection mode. Is also possible.

Example 9
In the ninth embodiment, as in the first embodiment, if an adjacent block is valid for an adjacent block on the same picture as the prediction block to be processed and a block adjacent to the same position of another picture having a different time, The information is registered in the reference candidate list, the optimum neighboring block coding information is selected from the reference candidate list, and the index of the reference candidate list representing the neighboring block is coded and transmitted.

  This will be described with reference to the arrangement of adjacent blocks in FIG. In the first embodiment, when creating a reference candidate list, a predetermined priority is associated with the position of an adjacent block, and it is determined whether the adjacent block satisfies a predetermined condition in the order of priority. When the condition is satisfied, the reference candidate list is created by registering in front of the reference candidate list. That is, the priority associated with the position of the adjacent block is reflected as it is in the index of the reference candidate list, and the higher the priority, the smaller the index is stored in the storage area of the reference candidate list.

  In general, it is considered that the neighboring blocks around the prediction block to be processed are more likely to have the same encoding information as the prediction block to be processed, as the area in contact with the prediction block to be processed is larger. By preferentially registering such adjacent blocks in front of the reference candidate list, the code amount of the index is reduced and the encoding efficiency is improved.

  From the above points, in the ninth embodiment, in the motion detection mode, the order of determining the validity of adjacent blocks when creating a reference candidate list for motion vector prediction and the order of codewords assigned to the reference adjacent blocks are separated. Defined in The order of determining the validity of adjacent blocks when creating the reference candidate list is such that the difference between the motion vectors of the left and upper adjacent block candidates becomes large in order to widen the selection of prediction motion vectors. Make the rankings separated by a distance.

  Here, B0, A0, A1, B1, C0, and T are used to determine the validity of adjacent blocks when creating a motion vector prediction reference candidate list. On the other hand, codewords assigned to reference neighboring blocks are assigned with a smaller number of codewords in the order of A1, B1, B0, A0, C0, and T, giving priority to neighboring blocks having a large area in contact with the processing target prediction block. I will do it.

  The motion vector predicting unit 103 according to the ninth embodiment installed in FIG. 2 illustrating the configuration of the moving image encoding device 100 will be described. The motion vector prediction unit 103 of the ninth embodiment has the same configuration as that of FIG. 23 described in the first embodiment, but the operation of the reference candidate list creation unit 321 is different. The operation of the reference candidate list creation unit 321 in the ninth embodiment will be described. FIG. 60 is a flowchart illustrating the operation of the reference candidate list creation unit 321 in the motion detection mode according to the ninth embodiment, which will be described with reference to this diagram.

  The reference candidate list creation unit 321 basically performs the same operation as the reference candidate list creation unit 321 in the motion vector prediction unit 103. In the motion detection mode, it is necessary to create a reference candidate list for each of the L0 prediction and the L1 prediction based on the reference list. Therefore, in the reference candidate list creation in the motion detection mode, the L0 prediction and the L1 prediction are separately performed twice. For this reason, in step S601 for determining the validity of the adjacent block, the flag availableFlagLXN (X is 0 or 1 and N is a symbol indicating the adjacent block) indicating the reference list is indicated in the LX indicating the reference block. Subscripts are added, but the determination operation is the same as in the merge mode.

In FIG. 60, creation / update (S608) of a conversion table is newly added to the flowchart of FIG. 19 showing the operation of the reference candidate list creation unit 321 in the motion vector prediction unit 103. This determination unit The operation of the reference candidate list creation unit 321 will be described. However, only one of the reference lists will be described here, and the determination operation is the same for the other reference list, and the description thereof will be omitted.

  First, the variable N and the index k of the reference candidate list are initialized (S600). In the variable N, the adjacent block B0 shown in FIG. 9A is initially set, and k is set to 0. The index k indicates the priority order of the storage areas for the encoded information candidates set in the storage area for the reference candidate list provided in the reference candidate list storage memory 324. This storage area has an array structure and is represented by candList. The index number starts from 0, and encoding information candidates of reference adjacent blocks are stored in the storage area of the reference candidate list candList. In the subsequent processing, the encoding information of the index i registered in the reference candidate list candList is represented by candList [i].

  Next, the valid flag availableFlagLXN of the reference adjacent block N is determined (S601). Here, 0 or 1 enters X. When availableFlagLXN is 0 (N in S601), it is not registered in the reference candidate list, and proceeds to the next. It is determined whether or not the adjacent block N is the last reference block (S604). If it is the last block (Y in S604), the value of the index k is set to the total number of candidate lists NumListCand (S605), and the process ends. If it is not the last block (N in S604), the variable N is updated (S606), and the processes after step S601 are repeated. Here, the order of adjacent blocks in which the variable N is updated is the order of priority for storage in the reference candidate list. In the ninth embodiment, the order (B0, A0, A1, B1, C0, T) is set. To do.

  On the other hand, if availableFlagLXN is 1 (Y in S601), the encoding information of the adjacent block N is registered in the reference candidate list candList [k] (S602), the conversion table is created / updated (S608), and k is set. Update (S603).

  Details of the creation / update (S608) of the conversion table will be described here. In the creation / update of the conversion table, a conversion table is created that associates the order of determining the validity of adjacent blocks with the reference candidate list for motion vector prediction and the order of codewords assigned to the reference adjacent blocks. The conversion table is a storage area having an array structure provided in the reference candidate list storage memory 324, and stores a conversion table index representing a codeword corresponding to the index of the reference candidate list.

  This will be described with reference to FIG. The variable N and the index k of the reference candidate list are input, and the creation / update of the conversion table starts. First, the variable M and the index j of the conversion table are initialized (S1400). The adjacent block A1 shown in FIG. 9A is initially set in the variable M, and j is set to 0. The variable M is updated in the order of A1, B1, B0, A0, C0, T. Next, the same determination of M and N is performed (S1401). When M is different from N (N in S1401), j is updated without being registered in the conversion table (S1403).

  Next, it is determined whether or not the variable M is the last adjacent block (S1404). When the last adjacent block is T and M is T (Y in S1404), the process ends. If it is not the last block (N in S1404), the variable M is updated (S1405), and the processes after step S1401 are repeated.

  When M is N (Y in S1401), the index k of the reference candidate list is registered in the jth storage area from the top of the conversion table (S1402). When the index k is registered, the process is terminated, and the process returns to the update of k in FIG. 60 (S603). Through the above processing, the index k of the reference candidate list is registered in the conversion table, and if all the reference neighboring blocks are valid, they are associated with the codeword represented by the index of the conversion table shown in FIG.

  If all the reference neighboring blocks are valid using the conversion table, the reference candidate list is created in the order shown in FIG. 63 and codewords are assigned. When only one reference adjacent block is valid, a codeword is not required, and a candidate of encoding information of an adjacent block determined to be only one valid is uniquely determined as a reference destination.

  Here, the index of the reference candidate list immediately after being created by the reference candidate list creating unit 321 has been described, but originally the adjacent block of the reference candidate having the same encoded information is identically detected by the same information detecting unit 322. It is desirable to associate the index of the reference candidate list with the index of the conversion table after deleting from the reference candidate list. The storage area of the conversion table corresponding to the index of the reference candidate list of the adjacent block of the reference candidate deleted by the same information detection unit 322 is also deleted and packed in the head direction of the conversion table.

  For example, in FIG. 62, when an adjacent block corresponding to index 1 in the reference candidate list is deleted, the location of index 3 in the conversion table corresponding to index 1 in the reference candidate list is deleted. The indexes 4 and 5 of the conversion table are 3 and 4, respectively, and the codewords associated with them are also updated to 1110 and 1111, respectively.

  As described above, the motion vector of the created reference candidate list is used as a predicted motion vector, and the difference motion vector calculation unit 326 calculates a difference motion vector from the motion vector detected by the motion vector detection unit 102 and the predicted motion vector. The calculated difference motion vector is supplied to the output unit 323. The output unit 323 outputs the index and the difference motion vector in the created reference candidate list. The reference candidate list outputs the reference candidate list as an MVP list and the index in the list as an MVP index.

  The motion vector calculation unit 204 installed in the video decoding device 200 corresponding to the motion vector prediction unit 103 installed in the video encoding device 100 is the same as the motion vector prediction unit 103. Since the reference candidate list creation unit 321 or the same information detection unit 322 according to the ninth embodiment having a function is provided, the same reference candidate list as the motion vector prediction unit 103 is created. The created reference candidate list is output to the selection unit 325.

  The selection unit 325 selects an adjacent block in the reference candidate list specified by an index for specifying the reference adjacent block decoded by the first encoded bit string decoding unit 202 from the created reference candidate list. A motion vector is output as a predicted motion vector from the encoded information of adjacent blocks in the selected list, and the motion vector predictor and the difference motion vector decoded by the first encoded bit string decoding unit 202 by the difference motion vector calculation unit 326 And the motion vector is calculated and supplied to the motion compensation prediction unit 207, and the encoded information of the adjacent block in the selected reference candidate list is supplied to the encoded information storage memory 210.

(Example 10)
In the tenth embodiment, in the motion detection mode, the order of determining the validity of adjacent blocks when creating the reference candidate list is changed, and the order of determining the subsequent adjacent blocks is changed depending on whether or not the first adjacent block is valid. Is. This will be described with reference to the arrangement of adjacent blocks in FIG. The order B0, A0, A1, C0, B1, and T for determining the validity of adjacent blocks when creating the reference candidate list according to the ninth embodiment will be described as an example.

  The adjacent block to be determined first is B0. When B0 is valid, the order of the subsequent adjacent blocks is not changed. When B0 is invalid, the order of subsequent adjacent blocks is changed to the order of A0, B1, C0, and A1. When B0 becomes invalid, A0 is the adjacent block to be determined first, and the next adjacent block is A1. Since A1 is close to A0, there is a high possibility that the motion vector has a value close to A0, and the possibility that A1 is selected as the predicted motion vector is low.

  The motion detection mode is a mode for transmitting a differential motion vector. When the blocks provided in the reference adjacent blocks are close to each other, the selection range of the prediction motion vector is narrowed, and an appropriate prediction motion vector cannot be detected. . Therefore, when B0 becomes invalid, the adjacent block to be determined next to A0 is separated from A0 by a distance so that the difference between A0 and the motion vector is increased in order to widen the selection of the predicted motion vector. Adjacent block B1 is selected. In this way, the order of determination is changed so that the distance between adjacent blocks that are candidates for the motion vector predictor is as far as possible.

(Example 11)
In the eleventh embodiment, in the motion detection mode, an adjacent block on the same picture as the prediction block to be processed is defined as a block group on the same picture, and two adjacent blocks representing the block group are selected from the adjacent blocks. One method is shown. This will be described with reference to the arrangement of adjacent blocks in FIG. Assume that the order of determining the validity of adjacent blocks when creating the reference candidate list is B0, B1, C0, A1, A0.

  First, the validity of adjacent blocks is determined in the listed left-to-right order B0, B1, C0, A1, A0 (forward direction), and the first adjacent block determined to be effective is the first block group. Is registered in the reference candidate list as an adjacent block representing. Next, the validity of the adjacent block is judged in the reverse order of the listed order, that is, the order from right to left A0, A1, C0, B1, B0 (reverse direction), and is referred to as an adjacent block representing another block group. Register in the candidate list.

  The order of determining the effectiveness of adjacent blocks is reversed from the upper right corner to the lower left corner of the prediction block to be processed so that the difference in motion vector of the adjacent block candidates is increased, and the candidate is reversed. By registering them with a distance between them, the selection range of the motion vector predictor is widened, the code amount of the difference motion vector is reduced, and the coding efficiency is improved.

  When determining the validity of the adjacent block in the forward direction and the reverse direction, the determination of the adjacent block corresponding to the last order of A0 in the forward direction and B0 in the reverse direction may be omitted. This is because when the last A0 is selected in the determination from the forward direction, it is an appropriate adjacent block selected in the determination from the forward direction because it is an adjacent block that is valid in the first determination from the reverse direction. .

  Further, the determination from each direction may be limited so that only three adjacent blocks up to C0 are determined. For example, when A0 is selected in the determination from the forward direction and A1 is selected in the determination from the reverse direction, there is a high possibility that the block is an adjacent block having the same motion vector, and the selection range of motion vector prediction is narrowed. An effective prediction motion vector cannot be derived. In order to avoid such selection of adjacent blocks adjacent to each other, the validity is determined up to C0 at the intermediate position. The limitation of validity determination as described above leads to a reduction in processing man-hours because the number of adjacent blocks to be determined is reduced.

(Example 12)
In the motion detection mode, the twelfth embodiment defines an adjacent block on the same picture as the prediction block to be processed as a block group on the same picture, and represents a block group among them in the motion detection mode. Are selected as adjacent blocks. In the twelfth embodiment, as in the ninth embodiment, the order of determining the validity of adjacent blocks when creating a motion vector prediction reference candidate list and the order of codewords assigned to the reference adjacent blocks are defined separately.

  When creating a reference candidate list, in order to widen the selection of predicted motion vectors, the distance between adjacent blocks should be separated so that the difference between the motion vectors of the left and upper adjacent blocks becomes large, and the codeword In the allocation, a small codeword is allocated preferentially to an adjacent block having a large area in contact with a processing target prediction block that is considered to be most likely to have the same encoding information as the processing target prediction block.

  Here, it is assumed that the order of determining the validity of adjacent blocks when creating a motion vector prediction reference candidate list is B0, A0, A1, C0, B1, T. On the other hand, codewords assigned to reference neighboring blocks are assigned with a smaller number of codewords in the order of A1, B1, B0, A0, C0, and T, giving priority to neighboring blocks having a large area in contact with the processing target prediction block. I will do it.

  First, the motion vector predicting unit 103, which is a feature of the twelfth embodiment installed in FIG. 2 showing the configuration of the video encoding device 100, will be described. The motion vector prediction unit 103 has the configuration shown in FIG. 64, and differs from FIG. 23 described in the ninth embodiment in that a reference candidate control unit 329 is added immediately after the reference candidate list creation unit 321.

  In the twelfth embodiment, since the adjacent block representing the block group on the same picture as the processing target prediction block is selected, the reference candidate list creation unit 321 is the same as the processing target prediction block described in the eighth embodiment. The same operation as the selection method from the block group on the picture is performed. Here, K representative neighboring blocks are selected from a block group on the same picture. The reference candidate list using the encoding information of K adjacent blocks representing the selected block group on the same picture and K + 1 adjacent blocks of the block T adjacent to the same position of another picture having different times as reference candidates Register with.

  Here, K is set to 2, and in the reference candidate list creation unit 321, two adjacent blocks representing the block group on the same picture from the top of the reference candidate list, and blocks adjacent to the same position of another picture having different times The following description will be given on the assumption that a total of three pieces of encoded information are registered in the reference candidate list.

  FIG. 65 is a flowchart showing the operation of the reference candidate control unit 329. The reference candidate list created by the reference candidate list creation unit 321 and the number of reference candidates NumListCand registered in the reference candidate list are input. First, NumListCand is determined (S1500). If NumListCand is less than 2 (Y in S1500), that is, if the reference candidate registered in the reference candidate list is 1, that one candidate is uniquely determined as the reference destination, and thus the process ends.

  If NumListCand is greater than 2 (N in S1500), N is the adjacent block including the encoded information CandListLX [1] registered second from the top of the reference candidate list. Here, since the reference candidate list array CandListLX starts with 0 as the head index of the reference candidate list, CandListLX [1] is a storage area for storing the second encoded information registered from the head of the reference candidate list. Subscript X contains 0 (L0 prediction) or 1 (L1 prediction) indicating the reference list of the reference list.

  Next, it is determined whether N is an adjacent block A1 or B1 (S1502). If N is A1 or not B1 (N in S1502), the process is terminated. When N is A1 or B1 (Y in S1502), the encoding information in the reference candidate list is replaced (S1503). The first encoded information CandListLX [0] and the second encoded information CandListLX [1] in the reference candidate list are exchanged.

  Since the order in which the reference candidate list is created is B0, A0, A1, C0, and B1, when A1 or B1 is second in the reference candidate list, the leading adjacent block has a codeword length longer than A1 or B1. become longer. This is because the codeword assignment order of the index specifying the adjacent block is defined as A1, B1, B0, A0, C0, T, so that the codeword length is A1 <B1 <B0 <A0 <C0 <T. Because of that. Therefore, A1 or B1 is replaced with the head of the reference candidate list, the priority in the reference candidate list of the adjacent block of the index with a short codeword length is increased, and the code amount of the index is reduced.

  In addition, even when the order in the reference candidate list before replacement is A1, B1, replacement is performed, and B1 with a long code word becomes the head of the reference candidate list, but other adjacent blocks are referred to. Loss of encoding efficiency can be suppressed as compared with the case where the candidate list is the head. As described above, replacement in the reference candidate list is performed, and it is possible to allocate adjacent blocks that specify adjacent blocks with a small number of codewords at the top of the reference candidate list, thereby improving encoding efficiency. It becomes.

  As described above, after the same information detection unit 322, the motion vector of the created reference candidate list is used as a predicted motion vector, and the differential motion vector calculation unit 326 detects the motion vector and the predicted motion detected by the motion vector detection unit 102. A difference motion vector is calculated from the vector, and the calculated difference motion vector is supplied to the output unit 323. The output unit 323 outputs the index and the difference motion vector in the created reference candidate list. The reference candidate list outputs the reference candidate list as an MVP list and the index in the list as an MVP index.

  The motion vector calculation unit 204 installed in the video decoding device 200 corresponding to the motion vector prediction unit 103 installed in the video encoding device 100 is the same as the motion vector prediction unit 103. Since the reference candidate control unit 329 according to the twelfth embodiment having a function is installed, the same reference candidate list as that of the motion vector prediction unit 103 is created. The created reference candidate list is output to the selection unit 325. The selection unit 325 selects an adjacent block in the reference candidate list specified by an index for specifying the reference adjacent block decoded by the first encoded bit string decoding unit 202 from the created reference candidate list. A motion vector is output as a predicted motion vector from the encoded information of adjacent blocks in the selected list, and the motion vector predictor and the difference motion vector decoded by the first encoded bit string decoding unit 202 by the difference motion vector calculation unit 326 And the motion vector is calculated and supplied to the motion compensation prediction unit 207, and the encoded information of the adjacent block in the selected reference candidate list is supplied to the encoded information storage memory 210.

  As described above, in the embodiment of the present invention, the arrangement of adjacent neighboring blocks referred to by the motion vector prediction method and the merging method is integrated to reduce the temporary memory for storing the neighboring block information, and the common By determining candidate priorities for each prediction method from the converted reference adjacent block, the redundancy of the encoding efficiency of the motion vector prediction method and the merge method in inter-picture prediction is reduced and the processing efficiency at the time of decoding is reduced. Can be raised. Also, before performing motion compensation, the number of motion compensation processes can be reduced by determining the effectiveness of referring to the block from the prediction mode and position information of neighboring neighboring blocks to be referenced.

  More specific description will be given below. In order to reduce the maximum number of memory accesses that can be assumed, the amount of memory, and the amount of processing (processing time), as shown in FIG. 9A, one block is adjacent to the left side and one block is adjacent to the upper side. Narrow down candidates by limiting. By limiting the candidates in this way in advance, there is an effect of reducing the number of memory accesses, the amount of memory, and the amount of processing (processing time) without substantially reducing the encoding efficiency.

  In addition, the number of memory accesses, the amount of memory, and the amount of processing (processing time) can be reduced, and the code amount of the merge index and the MVP index can be reduced. In addition, by sharing the candidate positions in the merge mode and the motion detection mode, the number of memory accesses, the amount of memory, and the amount of processing (processing time) are further reduced. Furthermore, by storing candidates in the candidate list in an arrangement according to the respective characteristics of the merge mode and the motion detection mode, the code amount of the merge index is reduced in the merge mode, and the MVP index and the difference motion vector in the motion detection mode. The coding efficiency is improved by reducing the amount of codes.

  The present invention is not limited to the above-described embodiments, and includes, for example, a moving image encoding / decoding program for causing a computer to realize the functions of the above moving image encoding / decoding device. This moving image encoding / decoding program may be read from a recording medium and taken into a computer, or may be transmitted via a communication network and taken into a computer.

  The encoded bit stream of the moving image output from the moving image encoding apparatus of the embodiment described above has a specific data format so that it can be decoded according to the encoding method used in the embodiment. Thus, a moving picture decoding apparatus corresponding to the moving picture encoding apparatus can decode the encoded bit stream of this specific data format.

  When a wired or wireless network is used to exchange an encoded bit stream between the moving image encoding device and the moving image decoding device, the encoded bit stream is converted into a data format suitable for the transmission form of the communication path. May be transmitted. In that case, a moving image transmitting apparatus that converts the encoded bit stream output from the moving image encoding apparatus into encoded data in a data format suitable for the transmission form of the communication path and transmits the encoded data to the network, and the encoded data from the network There is provided a moving image receiving apparatus that receives and restores the encoded bit stream and supplies the encoded bit stream to the moving image decoding apparatus.

  The moving image transmitting apparatus transmits a memory that buffers an encoded bit stream output from the moving image encoding apparatus, a packet processing unit that packetizes the encoded bit stream, and packetized encoded data via a network. And a transmitting unit. The moving image receiving apparatus generates a coded stream by packetizing the received data, a receiving unit that receives the packetized coded data via a network, a memory that buffers the received coded data, and packet processing. And a packet processing unit provided to the video decoding device.

  The present invention has been described based on several embodiments. Those embodiments are exemplifications, and it will be understood by those skilled in the art that various modifications can be made to combinations of the respective components and processing processes, and such modifications are also within the scope of the present invention. is there.

  DESCRIPTION OF SYMBOLS 100 moving image encoder, 101 image memory, 102 motion vector detection part, 103 motion vector prediction part, 104 inter prediction information detection part, 105 motion compensation prediction part, 106 merge detection part, 107 prediction method determination part, 108 switch, 109 first encoded bit string generation unit, 110 residual signal generation unit, 111 orthogonal transform / quantization unit, 112 inverse quantization / inverse orthogonal transform unit, 113 decoded image signal superimposition unit, 114 encoded information storage memory, 115 Decoded image memory, 116 second encoded bit string generation unit, 117 encoded bit string multiplexing unit, 200 moving image decoding apparatus, 201 bit string separation unit, 202 first encoded bit string decoding unit, 203 second encoded bit string Decoding unit, 204 motion vector calculation unit, 205 in Data prediction information detection unit, 206 merge determination unit, 207 motion compensation prediction unit, 208 inverse quantization / inverse orthogonal transform unit, 209 decoded image signal superposition unit, 210 encoded information storage memory, 211 decoded image memory, 212 switch, 301 Processing target block position detection unit, 302 adjacent block designation unit, 303 temporary memory, 310 encoded information derivation unit, 311 reference candidate list creation unit, 312 identical information detection unit, 313 output unit, 314 reference candidate list storage memory, 315 selection 318 reference candidate restriction unit 320 encoded information derivation unit 321 reference candidate list creation unit 322 identical information detection unit 323 output unit 324 reference candidate list storage memory 326 differential motion vector calculation unit 328 reference candidate restriction See 329 Auxiliary control unit.

Claims (6)

An image decoding device that decodes a code string encoded using a motion vector in units of blocks obtained by dividing each picture of a moving image,
For each decoding target block from the code string, prediction mode information indicating an inter-picture prediction mode and an index of a reference block, or prediction mode information indicating an inter-picture prediction mode, an index of a reference block, and a differential motion vector corresponding to the reference block, A code string decoding unit for decoding
An acquisition unit that acquires encoding information of a plurality of reference block candidates of the decoding target block;
When the inter-picture prediction mode indicated by the prediction mode information decoded by the code string decoding unit is the first inter-picture prediction mode using the motion vector information included in the coding information of the reference block specified by the index In addition, a first candidate list is generated from the plurality of reference block candidates based on a first predetermined order, and the first candidate list is generated based on the index of the reference block decoded by the code string decoding unit. A first mode output unit that identifies a reference block of the block to be decoded and outputs the encoded information;
When the inter-picture prediction mode indicated by the prediction mode information decoded by the code string decoding unit is the second inter-picture prediction mode using the differential motion vector corresponding to the reference block specified by the index, A second candidate list is generated from the plurality of reference block candidates based on a predetermined order, and based on the index of the reference block decoded by the code string decoding unit, the decoding target block from the second candidate list And a motion vector of the decoding target block is calculated from a prediction motion vector based on motion vector information included in the encoding information of the reference block and a difference motion vector decoded by the code string decoding unit. A second mode output unit for outputting
Based on the information output from the first mode output unit or the second mode output unit, motion compensation is performed using the inter-picture prediction mode indicated by the prediction mode information decoded by the code string decoding unit. A motion compensation prediction unit that generates a predicted image,
The first mode output unit and the second mode output unit use a plurality of common blocks as the plurality of reference block candidates,
The plurality of reference block candidates include an already decoded block spatially adjacent to the decoding target block and a block included in a picture temporally different from a picture including the decoding target block,
The first mode output unit and the second mode output unit are, in the plurality of reference block candidates, ranks of blocks included in the temporally different pictures in the first predetermined rank and the second predetermined rank, respectively. Is set lower than the ranks in the first predetermined rank and the second predetermined rank of the spatially adjacent blocks.
An image decoding apparatus characterized by that. An image decoding method for decoding a code string encoded using a motion vector in units of blocks obtained by dividing each picture of a moving image,
For each decoding target block from the code string, prediction mode information indicating an inter-picture prediction mode and an index of a reference block, or prediction mode information indicating an inter-picture prediction mode, an index of a reference block, and a differential motion vector corresponding to the reference block, A first step of decoding
A second step of obtaining encoding information of a plurality of reference block candidates of the decoding target block;
When the inter-picture prediction mode indicated by the prediction mode information decoded in the first step is the first inter-picture prediction mode that uses information on motion vectors included in the encoded information of the reference block specified by the index. , Generating a first candidate list from the plurality of reference block candidates based on a first predetermined order, and from the first candidate list based on the index of the reference block decoded in the first step A third step of identifying a reference block of a decoding target block and outputting the encoding information;
When the inter-picture prediction mode indicated by the prediction mode information decoded in the first step is the second inter-picture prediction mode using the differential motion vector corresponding to the reference block specified by the index, the second predetermined Generating a second candidate list from the plurality of reference block candidates based on the ranking, and referring to the decoding target block from the second candidate list based on the index of the reference block decoded in the first step A block is specified, and a motion vector of the decoding target block is calculated and output from a motion vector predictor based on motion vector information included in the encoding information of the reference block and the differential motion vector decoded in the first step. The fourth step;
Based on the information output in the third step or the fourth step, a predicted image is generated by performing motion compensation using the inter-picture prediction mode indicated by the prediction mode information decoded in the first step. And comprising steps
The third step and the fourth step use a plurality of common blocks as the plurality of reference block candidates,
The plurality of reference block candidates include an already decoded block spatially adjacent to the decoding target block and a block included in a picture temporally different from a picture including the decoding target block,
In the third step and the fourth step, in the plurality of reference block candidates, the ranks of the blocks included in the temporally different pictures in the first predetermined rank and the second predetermined rank are spatially Set lower than the ranks in the first predetermined rank and the second predetermined rank of adjacent blocks.
An image decoding method characterized by the above. An image decoding program for decoding a code string encoded using a motion vector in units of blocks obtained by dividing each picture of a moving image,
For each decoding target block from the code string, prediction mode information indicating an inter-picture prediction mode and an index of a reference block, or prediction mode information indicating an inter-picture prediction mode, an index of a reference block, and a differential motion vector corresponding to the reference block, A first process for decoding
A second process of acquiring encoding information of a plurality of reference block candidates of the decoding target block;
When the inter-picture prediction mode indicated by the prediction mode information decoded by the first process is the first inter-picture prediction mode that uses information on motion vectors included in the encoded information of the reference block specified by the index. , Generating a first candidate list from the plurality of reference block candidates based on a first predetermined order, and from the first candidate list based on the index of the reference block decoded by the first process A third process for identifying a reference block of a decoding target block and outputting the encoded information;
When the inter-picture prediction mode indicated by the prediction mode information decoded by the first process is the second inter-picture prediction mode using the differential motion vector corresponding to the reference block specified by the index, the second predetermined Generating a second candidate list from the plurality of reference block candidates based on the ranking, and referring to the decoding target block from the second candidate list based on an index of the reference block decoded by the first process A block is specified, and a motion vector of the decoding target block is calculated and output from a predicted motion vector based on motion vector information included in the encoded information of the reference block and a differential motion vector decoded by the first process. A fourth process;
Based on the information output by the third process or the fourth process, a prediction image is generated by performing motion compensation using the inter-picture prediction mode indicated by the prediction mode information decoded by the first process. Processing to the computer,
The third process and the fourth process use a plurality of common blocks as the plurality of reference block candidates,
The plurality of reference block candidates include an already decoded block spatially adjacent to the decoding target block and a block included in a picture temporally different from a picture including the decoding target block,
In the third process and the fourth process, in the plurality of reference block candidates, the ranks of the blocks included in the temporally different pictures in the first predetermined rank and the second predetermined rank are spatially Set lower than the ranks in the first predetermined rank and the second predetermined rank of adjacent blocks.
An image decoding program characterized by the above. A receiving device that receives and decodes a code string in which a moving image is encoded,
A receiving unit that receives an encoded stream in which a code sequence encoded using a motion vector in units of blocks obtained by dividing each picture of a moving image is packetized;
A restoration unit that packet-processes the received packetized coded stream to restore the original code string;
Corresponding to prediction mode information indicating inter-picture prediction mode and reference block index, or prediction mode information indicating inter-picture prediction mode, reference block index and reference block for each decoding target block from the restored original code string A code string decoding unit for decoding the difference motion vector obtained,
An acquisition unit that acquires encoding information of a plurality of reference block candidates of the decoding target block;
When the inter-picture prediction mode indicated by the prediction mode information decoded by the code string decoding unit is the first inter-picture prediction mode using the motion vector information included in the coding information of the reference block specified by the index In addition, a first candidate list is generated from the plurality of reference block candidates based on a first predetermined order, and the first candidate list is generated based on the index of the reference block decoded by the code string decoding unit. A first mode output unit that identifies a reference block of the block to be decoded and outputs the encoded information;
When the inter-picture prediction mode indicated by the prediction mode information decoded by the code string decoding unit is the second inter-picture prediction mode using the differential motion vector corresponding to the reference block specified by the index, A second candidate list is generated from the plurality of reference block candidates based on a predetermined order, and based on the index of the reference block decoded by the code string decoding unit, the decoding target block from the second candidate list And a motion vector of the decoding target block is calculated from a prediction motion vector based on motion vector information included in the encoding information of the reference block and a difference motion vector decoded by the code string decoding unit. A second mode output unit for outputting
Based on the information output from the first mode output unit or the second mode output unit, motion compensation is performed using the inter-picture prediction mode indicated by the prediction mode information decoded by the code string decoding unit. A motion compensation prediction unit that generates a predicted image,
The first mode output unit and the second mode output unit use a plurality of common blocks as the plurality of reference block candidates,
The plurality of reference block candidates include an already decoded block spatially adjacent to the decoding target block and a block included in a picture temporally different from a picture including the decoding target block,
The first mode output unit and the second mode output unit are, in the plurality of reference block candidates, ranks of blocks included in the temporally different pictures in the first predetermined rank and the second predetermined rank, respectively. Is set lower than the ranks in the first predetermined rank and the second predetermined rank of the spatially adjacent blocks.
A receiving apparatus. A receiving method for receiving and decoding a code string in which a moving image is encoded,
A reception step of receiving an encoded stream in which a code sequence encoded using a motion vector in units of blocks obtained by dividing each picture of a moving image is packetized;
A restoration step of packetizing the received packetized coded stream to restore the original code sequence;
Corresponding to prediction mode information indicating inter-picture prediction mode and reference block index, or prediction mode information indicating inter-picture prediction mode, reference block index and reference block for each decoding target block from the restored original code string A first step of decoding the difference motion vector obtained;
A second step of obtaining encoding information of a plurality of reference block candidates of the decoding target block;
When the inter-picture prediction mode indicated by the prediction mode information decoded in the first step is the first inter-picture prediction mode that uses information on motion vectors included in the encoded information of the reference block specified by the index. , Generating a first candidate list from the plurality of reference block candidates based on a first predetermined order, and from the first candidate list based on the index of the reference block decoded in the first step A third step of identifying a reference block of a decoding target block and outputting the encoding information;
When the inter-picture prediction mode indicated by the prediction mode information decoded in the first step is the second inter-picture prediction mode using the differential motion vector corresponding to the reference block specified by the index, the second predetermined Generating a second candidate list from the plurality of reference block candidates based on the ranking, and referring to the decoding target block from the second candidate list based on the index of the reference block decoded in the first step A block is specified, and a motion vector of the decoding target block is calculated and output from a motion vector predictor based on motion vector information included in the encoding information of the reference block and the differential motion vector decoded in the first step. The fourth step;
Based on the information output in the third step or the fourth step, a predicted image is generated by performing motion compensation using the inter-picture prediction mode indicated by the prediction mode information decoded in the first step. And comprising steps
The third step and the fourth step use a plurality of common blocks as the plurality of reference block candidates,
The plurality of reference block candidates include an already decoded block spatially adjacent to the decoding target block and a block included in a picture temporally different from a picture including the decoding target block,
In the third step and the fourth step, in the plurality of reference block candidates, the ranks of the blocks included in the temporally different pictures in the first predetermined rank and the second predetermined rank are spatially Set lower than the ranks in the first predetermined rank and the second predetermined rank of adjacent blocks.
And a receiving method. A receiving program for receiving and decoding a code string in which a moving image is encoded,
A receiving process for receiving an encoded stream in which a code string encoded using a motion vector in units of blocks obtained by dividing each picture of a moving image is packetized;
A restoration process for packetizing the received packetized encoded stream to restore the original code string;
Corresponding to prediction mode information indicating inter-picture prediction mode and reference block index, or prediction mode information indicating inter-picture prediction mode, reference block index and reference block for each decoding target block from the restored original code string A first process for decoding the difference motion vector,
A second process of acquiring encoding information of a plurality of reference block candidates of the decoding target block;
When the inter-picture prediction mode indicated by the prediction mode information decoded by the first process is the first inter-picture prediction mode that uses information on motion vectors included in the encoded information of the reference block specified by the index. , Generating a first candidate list from the plurality of reference block candidates based on a first predetermined order, and from the first candidate list based on the index of the reference block decoded by the first process A third process for identifying a reference block of a decoding target block and outputting the encoded information;
When the inter-picture prediction mode indicated by the prediction mode information decoded by the first process is the second inter-picture prediction mode using the differential motion vector corresponding to the reference block specified by the index, the second predetermined Generating a second candidate list from the plurality of reference block candidates based on the ranking, and referring to the decoding target block from the second candidate list based on an index of the reference block decoded by the first process A block is specified, and a motion vector of the decoding target block is calculated and output from a predicted motion vector based on motion vector information included in the encoded information of the reference block and a differential motion vector decoded by the first process. A fourth process;
Based on the information output by the third process or the fourth process, a prediction image is generated by performing motion compensation using the inter-picture prediction mode indicated by the prediction mode information decoded by the first process. Processing to the computer,
The third process and the fourth process use a plurality of common blocks as the plurality of reference block candidates,
The plurality of reference block candidates include an already decoded block spatially adjacent to the decoding target block and a block included in a picture temporally different from a picture including the decoding target block,
In the third process and the fourth process, in the plurality of reference block candidates, the ranks of the blocks included in the temporally different pictures in the first predetermined rank and the second predetermined rank are spatially Set lower than the ranks in the first predetermined rank and the second predetermined rank of adjacent blocks.
A receiving program characterized by the above.

Images