BR122020016747B1 - MOTION REPRESENTATION DECODING DEVICE - Google Patents

Abstract

MOTION REPRESENTATION DECODING DEVICE. Confluent motion information candidate list building unit (162) generates a candidate list using spatial confluent motion information candidate and temporal confluent motion information candidate. Second confluent movement information candidate supply unit (165) generates new candidate including, if a reference index indicates an available reference representation, the reference index and, if, the reference index does not indicate an available reference representation , a predetermined reference index along with a motion vector having a preset size and direction, and adds the new candidate to the candidate list of confluent motion information. A confluent movement information selection unit (141) selects a confluent movement information candidate from the list to which the new candidate has been added and derives the selected candidate as movement information from the prediction block subjected to encoding.

Description

FIELD OF TECHNIQUE

[001] The present invention relates to a motion representation encoding technology using motion compensation prediction, and more particularly to a motion representation encoding device, a motion representation encoding method and a motion representation encoding program and a motion representation decoding device, a motion representation decoding method and a motion representation decoding program encoding motion information used in motion compensation prediction.

BACKGROUND OF THE INVENTION

[002] Motion compensation prediction is used in commonly used motion representation compression encoding. Motion compensation prediction is a technology of partitioning a target representation into small blocks and generating, as a prediction signal, a signal situated at a moved position from a target block of the target representation to a reference block of a reference representation based on the amount of motion indicated by a motion vector, where the reference representation is a decoded representation. Motion compensation prediction includes a uni-prediction performed using a single motion vector and a dual prediction performed using two motion vectors.

[003] With regard to motion vectors, a motion vector of a coded block adjacent to the target block is defined to be a motion vector prediction (also simply referred to as "vector prediction") and a difference between a motion vector target block movement and vector prediction is obtained and transmitted as a coding vector in order to improve the compression efficiency.

[004] Motion representation compression encoding such as MPEG-4AVC/H.264 (hereinafter AVC) enables highly accurate motion compensation prediction by making the size of a block used for motion compensation of movement is small with variations. On the other hand, there is a problem that the code size of an encoding array becomes large when reducing the block size.

[005] Thus, in AVC, the continuity of motion in a temporal direction is focused and the prediction of direct temporal motion compensation is used in which motion compensation is achieved, without transmitting the coding vectors, scaling in scales a motion vector of a block of a reference representation that is situated at the same position as a target block and using the scaled motion vector as a motion vector of the target block.

[006] Also, focusing on the continuity of motion in a spatial direction, Patent Document 1 discloses a motion compensation prediction range method, without transmitting coding vectors, using a motion vector of a block processed adjacent to the target block as a motion vector of the processed block.

RELATED TECHNIQUE LIST PATENT DOCUMENT

[007] PATENT DOCUMENT 1 JP 10-276439

DESCRIPTION OF THE INVENTION

[008] According to the method described in AVC or in the non-patent document 1, only one prediction block and one right mode can be obtained. Depending on the type of a representation, this can cause poor prediction accuracy provided by a motion vector prediction, which results in poor coding efficiency.

[009] Against this background, the inventors of the present invention became aware of a need for providing a motion representation encoding scheme based on motion compensation prediction in which the encoding information is compressed and the overall code size is reduced.

[0010] Accordingly, a purpose of the present invention is to provide a motion representation encoding and decoding technology capable of reducing, by providing a plurality of motion information candidates when motion information is not transmitted, the code size motion information and consequently improve coding efficiency.

[0011] A motion representation encoding device according to an embodiment of the present invention is a motion representation encoding device adapted to derive motion information including a reference index and a motion vector in block units and includes: a spatial confluent motion information candidate derivation unit (160) configured to derive a spatial confluent motion information candidate from the motion information of a plurality of coded prediction blocks adjacent to a prediction block. prediction submitted to coding; a temporal confluent motion information candidate derivation unit (161) configured to derive a temporal confluent motion information candidate from the motion information of a prediction block in a coded representation other than a representation in which the prediction block prediction submitted to coding is situated; a confluent motion information candidate list building unit (162) configured to generate a confluent motion information candidate list, which is a confluent motion information candidate list, using the confluent motion information candidate spatial and temporal confluent motion information candidate; a confluent motion information candidate supply unit (165) configured to generate a new confluent motion information candidate that includes, if the reference index indicates an available reference representation, the reference index and, if the reference index reference does not indicate an available reference representation, a predetermined reference index along with a motion vector that has a size and direction that are preset, and add the new confluent motion information candidate to the motion information candidate list confluent; a confluent movement information selection unit (141) configured to select a confluent movement information candidate from the confluent movement information candidate list to which the new confluent movement information candidate has been added and derive the candidate from selected confluent motion information as the motion information of the prediction block subjected to encoding; and an encoding unit (104) configured to encode a specification index for specifying the selected confluent motion information candidate.

[0012] Another embodiment of the present invention also relates to a moving representation coding device. This device is a motion representation encoding device adapted to derive motion information including a reference index and motion vector in units of prediction blocks and includes: a spatial confluent motion information candidate derivation unit ( 160) configured to derive a spatial confluent motion information candidate from the motion information of a plurality of encoded prediction blocks adjacent to a prediction block undergoing encoding; a temporal confluent motion information candidate derivation unit (161) configured to derive a temporal confluent motion information candidate from the motion information of a prediction block in a coded representation other than a representation in which the prediction block prediction submitted to coding is situated; a confluent motion information candidate list construction unit (162) configured to generate a confluent motion information candidate list, which is a motion information candidate list, using the spatial confluent motion information candidate and the temporal confluent movement information candidate; a first confluent motion information candidate supply unit (164) configured to combine a motion vector and a reference index of a confluent motion information candidate for the first prediction and a motion vector and a reference index of a confluent motion information candidate for the second prediction from the confluent motion information candidate list, derive a first supplementary confluent motion information candidate in which a first prediction prediction signal and a second prediction prediction signal are not identical to each other and add the first supplemental confluent motion information candidate to the list of confluent motion information candidates; a second confluent motion information candidate supply unit (165) configured to derive a second supplementary confluent motion information candidate having a motion vector and a reference index that allows the prediction signal of the first prediction and the prediction signal of the second prediction are identical to each other, and add the second supplemental confluent motion information candidate to the list of confluent motion information candidates; a confluent movement information selection unit (141) configured to select a confluent movement information candidate from the list of confluent movement information candidates which the first confluent movement information candidate supplements and the second confluent movement information candidate supplemental confluent motion were added and to derive the selected confluent motion information candidate as the motion information of the prediction block subjected to encoding; and an encoding unit (104) configured to encode a specification index for specifying the selected confluent motion information candidate.

[0013] Yet another embodiment of the present invention also relates to a moving representation coding device. That device is a motion representation encoding device adapted for partitioning a representation into a plurality of prediction blocks and performing motion compensation, and includes: a spatial confluent motion information candidate derivation unit (160) configured to deriving a spatially confluent motion information candidate from the motion information of a plurality of coded adjacent blocks adjacent to a prediction block undergoing encoding; a temporal confluent motion information candidate derivation unit (161) configured to derive a temporal confluent motion information candidate from the motion information of a block in a coded representation other than a representation subjected to encoding in which the block of prediction subjected to coding is situated; a confluent motion information candidate list construction unit (162) configured to generate a confluent motion information candidate list using the spatial confluent motion information candidate and the temporal confluent motion information candidate; a confluent motion information candidate supply unit (165) configured to generate a confluent motion information candidate for dual prediction where a motion vector value is the same for the first prediction and the second prediction and where an index of reference of the first prediction or the second prediction is set to be a fixed value and add the confluent motion information candidate to the confluent motion information candidate list; a confluent movement information selection unit (141) configured to select a confluent movement information candidate from the confluent movement information candidate list and define the selected confluent movement information candidate to be the movement information of the prediction block subjected to encoding; and an encoding unit (104) configured to encode information to specify the confluent motion information candidate selected from the confluent motion information candidate list.

[0014] Yet another embodiment of the present invention relates to a moving representation encoding method. This method is a motion representation encoding method adapted for deriving motion information that includes a reference index and motion vector in units of prediction blocks, and includes: deriving a spatial confluent motion information candidate from the information moving a plurality of encoded prediction blocks adjacent to a prediction block undergoing encoding; deriving a temporally confluent motion information candidate from the motion information of a prediction block in an encoded representation other than a representation in which the prediction block undergoing encoding is situated; generating a confluent motion information candidate list, which is a confluent motion information candidate list, using the spatial confluent motion information candidate and the temporal confluent motion information candidate; generate a new confluent movement information candidate that includes, if the reference index indicates an available reference representation, the reference index and, if the reference index does not indicate an available reference representation, a predetermined reference index along with a motion vector having a size and direction that are preset and which generates the new confluent motion information candidate to the list of confluent motion information candidates; select a confluent motion information candidate from the list of confluent motion information candidates to which the new confluent motion information candidate has been added and which derives the selected confluent motion information candidate as the motion information of the prediction submitted to coding; and encoding a specification index to specify the selected confluent motion information candidate.

[0015] Yet another embodiment of the present invention also relates to a moving representation encoding method. This method is a motion representation encoding method adapted for deriving motion information that includes a reference index and motion vector in units of prediction blocks, and includes: deriving a spatial confluent motion information candidate from the information moving a plurality of encoded prediction blocks adjacent to a prediction block undergoing encoding; deriving a temporally confluent motion information candidate from the motion information of a prediction block in an encoded representation other than a representation in which the prediction block undergoing encoding is situated; generating a confluent motion information candidate list, which is a motion information candidate list, using the spatial confluent motion information candidate and the temporal confluent motion information candidate; combine a motion vector and a reference index of a confluent motion information candidate for the first prediction and a motion vector and a reference index of a confluent motion information candidate for the second prediction from the candidate list of confluent motion information, deriving a first candidate of supplementary confluent motion information in which a prediction signal of the first prediction and a prediction signal of the second prediction are not identical with each other, and which adds the first candidate of motion information confluent supplementary to confluent motion information candidate list; deriving a second candidate from supplemental confluent motion information that has a motion vector and a reference index that allows the prediction signal of the first prediction and the prediction signal of the second prediction to be identical to each other and that adds the second candidate of confluent movement information supplemental to the candidate list of confluent movement information; select a confluent motion information candidate from the list of confluent motion information candidates to which the first supplemental confluent motion information candidate and the second supplementary confluent motion information candidate have been added and which derives the confluent motion selected as the motion information of the prediction block subjected to encoding; and encoding a specification index to specify the selected confluent motion information candidate.

[0016] A motion representation decoding device according to an embodiment of the present invention is a motion representation decoding device adapted to derive motion information including a reference index and a motion vector in units of motion blocks prediction and includes: a decoding unit (201) configured to decode, from a bitstream, a specification index for specifying a confluent motion information candidate used for a prediction block undergoing decoding; a spatial confluent motion information candidate derivation unit (160) configured to derive a spatial confluent motion information candidate from the motion information of a plurality of decoded prediction blocks adjacent to the prediction block undergoing decoding; a temporal confluent motion information candidate derivation unit (161) configured to derive a temporal confluent motion information candidate from the motion information of a prediction block in a decoded representation other than a representation in which the prediction block preview submitted to decoding is situated; a confluent motion information candidate list building unit (162) configured to generate a confluent motion information candidate list, which is a confluent motion information candidate list, using the confluent motion information candidate spatial and temporal confluent motion information candidate; a confluent motion information candidate supply unit (165) configured to generate a new confluent motion information candidate that includes, if the reference index indicates an available reference representation, the reference index and, if the reference index reference does not indicate an available reference representation, a predetermined reference index along with a motion vector that has a size and direction that are preset, and add the new confluent motion information candidate to the motion information candidate list confluent; and a confluent motion information selection unit (231) configured to select a confluent motion information candidate from the list of confluent motion information candidates to which the new confluent motion information candidate has been added and derive the candidate of confluent motion information selected as the motion information of the prediction block subjected to decoding, based on the specification index that was decoded.

[0017] Another embodiment of the present invention also relates to a moving representation decoding device. That device is a motion representation decoding device adapted to derive motion information including a reference index and a motion vector in units of prediction blocks and includes: a decoding unit (201) configured to decode, from from a bitstream, a specification index for specifying a confluent motion information candidate used for a prediction block subjected to decoding; a spatial confluent motion information candidate derivation unit (160) configured to derive a spatial confluent motion information candidate from the motion information of a plurality of decoded prediction blocks adjacent to the prediction block undergoing decoding; a temporal confluent motion information candidate derivation unit (161) configured to derive a temporal confluent motion information candidate from the motion information of a prediction block in a decoded representation other than a representation in which the prediction block preview submitted to decoding is situated; a confluent motion information candidate list construction unit (162) configured to generate a confluent motion information candidate list, which is a motion information candidate list, using the spatial confluent motion information candidate and the temporal confluent movement information candidate; a first confluent motion information candidate supply unit (164) configured to combine a motion vector and a reference index of a confluent motion information candidate for the first prediction and a motion vector and a reference index of a confluent motion information candidate for the second prediction from the confluent motion information candidate list, derive a first supplementary confluent motion information candidate in which a first prediction prediction signal and a second prediction prediction signal are not identical to each other and add the first supplemental confluent motion information candidate to the list of confluent motion information candidates; a second confluent motion information candidate supply unit (165) configured to derive a second supplementary confluent motion information candidate having a motion vector and a reference index that allows the prediction signal of the first prediction and the prediction signal of the second prediction are identical to each other, and add the second supplemental confluent motion information candidate to the list of confluent motion information candidates; and a confluent motion information selection unit (231) configured to select a confluent motion information candidate from the list of confluent motion information candidates which the first confluent motion information candidate supplements and the second confluent motion information candidate Supplementary confluent motion information was added and derived the selected confluent motion information candidate as the motion information of the prediction block submitted for decoding, based on the specification index that was decoded.

[0018] Yet another embodiment of the present invention relates to a moving representation decoding device. That device is a motion representation decoding device adapted for partitioning a representation into a plurality of prediction blocks and performing motion compensation, and includes: a decoding unit (201) configured to decode, from a data stream, bits into which an index for specifying a candidate of confluent motion information used in a prediction block subjected to decoding is encoded, the index; a spatial confluent motion information candidate derivation unit (160) configured to derive a spatial confluent motion information candidate from the motion information of a plurality of decoded adjacent blocks adjacent to the prediction block undergoing decoding; a temporal confluent motion information candidate derivation unit (161) configured to derive a temporal confluent motion information candidate from the motion information of a block in a decoded representation other than a representation subjected to decoding in which the block of prediction subjected to decoding is located; a confluent motion information candidate list construction unit (162) configured to generate a confluent motion information candidate list using the spatial confluent motion information candidate and the temporal confluent motion information candidate; a confluent motion information candidate supply unit (165) configured to generate a confluent motion information candidate for dual prediction where a motion vector value is the same for the first prediction and the second prediction and where an index of reference of the first prediction or the second prediction is set to be a fixed value and add the confluent motion information candidate to the confluent motion information candidate list; and a confluent motion information selection unit (231) configured to select a confluent motion information candidate from the list of confluent motion information candidates based on the index that has been decoded and define the confluent motion information candidate selected to be the motion information of the prediction block submitted for decoding.

[0019] Yet another embodiment of the present invention relates to a moving representation decoding method. This method is a motion representation decoding method adapted for deriving motion information that includes a reference index and a motion vector in units of prediction blocks and includes: decoding, from a bit stream, an index of specification for specifying a confluent motion information candidate used for a prediction block undergoing decoding; deriving a spatial confluent motion information candidate from the motion information of a plurality of decoded prediction blocks adjacent to the prediction block undergoing decoding; deriving a temporally confluent motion information candidate from the motion information of a prediction block in a decoded representation other than a representation in which the prediction block undergoing decoding is situated; generating a confluent motion information candidate list, which is a confluent motion information candidate list, using the spatial confluent motion information candidate and the temporal confluent motion information candidate; generate a new confluent movement information candidate that includes, if the reference index indicates an available reference representation, the reference index and, if the reference index does not indicate an available reference representation, a predetermined reference index along with a motion vector having a size and direction that are preset and which generates the new confluent motion information candidate to the list of confluent motion information candidates; and selecting a confluent motion information candidate from the list of confluent motion information candidates to which the new confluent motion information candidate has been added, and deriving the selected confluent motion information candidate as the motion information of the prediction subjected to decoding, based on the specification index that was decoded.

[0020] Yet another embodiment of the present invention also relates to a moving representation decoding method. This method is a motion representation decoding method adapted for deriving motion information that includes a reference index and a motion vector in units of prediction blocks and includes: decoding, from a bit stream, an index of specification for specifying a confluent motion information candidate used for a prediction block undergoing decoding; deriving a spatial confluent motion information candidate from the motion information of a plurality of decoded prediction blocks adjacent to the prediction block undergoing decoding; deriving a temporally confluent motion information candidate from the motion information of a prediction block in a decoded representation other than a representation in which the prediction block undergoing decoding is situated; generating a confluent motion information candidate list, which is a motion information candidate list, using the spatial confluent motion information candidate and the temporal confluent motion information candidate; combine a motion vector and a reference index of a confluent motion information candidate for the first prediction and a motion vector and a reference index of a confluent motion information candidate for the second prediction from the candidate list of confluent motion information, deriving a first candidate of supplementary confluent motion information in which a prediction signal of the first prediction and a prediction signal of the second prediction are not identical with each other, and which adds the first candidate of motion information confluent supplementary to confluent motion information candidate list; deriving a second candidate from supplemental confluent motion information that has a motion vector and a reference index that allows the prediction signal of the first prediction and the prediction signal of the second prediction to be identical to each other and that adds the second candidate of confluent movement information supplemental to the candidate list of confluent movement information; and selecting a confluent motion information candidate from the list of confluent motion information candidates to which the first supplemental confluent motion information candidate and the second supplementary confluent motion information candidate have been added, and deriving the supplementary motion information candidate from confluent motion selected as the motion information of the prediction block subjected to decoding, based on the specification index that was decoded.

[0021] Optional combinations of the aforementioned constituent elements and implementations of the invention in the form of methods, apparatus, systems, recording media and computer programs can also be practiced as additional modes of the present invention.

[0022] According to the present invention, by providing a plurality of motion information candidates when motion information is not transmitted, a reduction can be achieved in the code size of the motion information. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figures 1A and 1B are diagrams that explain the coding blocks;

[0024] Figures 2A through 2D are diagrams explaining the prediction block size tops;

[0025] Figure 3 is a diagram explaining types of prediction block size;

[0026] Figure 4 is a diagram explaining the predictive encoding modes;

[0027] Figure 5 is a diagram explaining the relationships between confluence indices and bit streams;

[0028] Figure 6 is a diagram that explains a syntax example of a prediction block;

[0029] Figure 7 is a diagram showing the configuration of a moving representation coding device according to a first embodiment;

[0030] Figure 8 is a flowchart explaining the operation of building a L0 reference representation list;

[0031] Fig. 9 is a flowchart explaining the operation of building an L1 reference representation list;

[0032] Figure 10 is a diagram showing the configuration of a movement information generation unit shown in Figure 7;

[0033] Figure 11 is a diagram explaining the configuration of a confluence mode determination unit shown in Figure 10;

[0034] Figure 12 is a flowchart explaining the operation of the confluence mode determination unit;

[0035] Figure 13 is a diagram explaining the configuration of a confluent movement information candidate list building unit shown in Figure 11;

[0036] Figure 14 is a flowchart explaining the operation of the confluent movement information candidate list building unit shown in Figure 11;

[0037] Figures 15A and 15B are diagrams explaining a group of special candidate blocks of a prediction block;

[0038] Figure 16 is a flowchart explaining the operation of a spatial confluent motion information candidate derivation unit shown in Figure 13;

[0039] Figure 17 is a flowchart explaining the operation of a temporal confluent motion information candidate derivation unit shown in Figure 13;

[0040] Figure 18 is a flow chart explaining the operation of a first confluent motion candidate information supply unit shown in Figure 13;

[0041] Figure 19 is a diagram explaining the relationships between the amount of combination checks, a confluent movement information candidate M and a confluent movement information candidate N;

[0042] Figure 20 is a flow chart explaining the operation of a second confluent motion candidate information supply unit shown in Figure 13;

[0043] Figures 21A and 21B are diagrams explaining a relationship between a target representation and a reference representation;

[0044] Figures 22A and 22B are diagrams explaining an example of a reference representation list;

[0045] Figures 23A and 23B are diagrams explaining an example of a relationship between a reference index of a second candidate of supplemental confluent movement information and a POC according to the first embodiment;

[0046] Figure 24 is a diagram showing the configuration of a vector prediction mode determination unit;

[0047] Figure 25 is a flowchart showing the operation of the vector prediction mode determination unit;

[0048] Figure 26 is a diagram explaining the configuration of a vector prediction candidate list building unit;

[0049] Figure 27 is a flowchart explaining the operation of the vector prediction candidate list building unit;

[0050] Figure 28 is a flow chart explaining the operation of a space vector predictor candidate derivation unit;

[0051] Figure 29 is a flowchart explaining the operation of a space-scale scaling prediction vector prediction candidate derivation unit;

[0052] Fig. 30 is a flowchart explaining the operation of a temporal vector prediction candidate derivation unit;

[0053] Fig. 31 is a diagram showing the configuration of a moving representation decoding device according to the first embodiment;

[0054] Figure 32 is a diagram showing the configuration of a movement information reproduction unit shown in Figure 31;

[0055] Figure 33 is a diagram showing the configuration of a confluent motion information reproduction unit shown in Figure 32;

[0056] Figure 34 is a diagram explaining the operation of a confluent movement information reproduction unit;

[0057] Figure 35 is a diagram explaining the configuration of a motion vector reproduction unit;

[0058] Fig. 36 is a diagram explaining the operation of the motion vector reproduction unit;

[0059] Fig. 37 is a flowchart explaining the operation of a second confluent movement information candidate supply unit according to a second embodiment;

[0060] Figures 38A and 38B are diagrams explaining an example of a relationship between a reference index of a second supplemental confluent movement information candidate and a POC according to the second embodiment;

[0061] Figure 39 is a flowchart explaining the operation of a second confluent movement information candidate supply unit according to a third embodiment; It is

[0062] Figures 40A and 40B are diagrams explaining an example of a relationship between a reference index of a second candidate of supplemental confluent movement information and a POC according to the third embodiment. BEST MODE TO PERFORM THE INVENTION

[0063] First, an explanation is given of a technology on which the embodiments of the present invention are based.

[0064] Currently, devices and systems obey a coding system such as, for example, MPEG (Group of Experts in Moving Image) or similar have become widely used. In such an encoding system, a plurality of representations that are continuous on a time axis are treated as digital signal information. In this case, for the purpose of highly efficient diffusion, transmission, accumulation and the like of information, compression encoding is performed using motion compensation prediction in which a representation is partitioned into a plurality of blocks and the redundancy into a temporal direction is used and with the use of orthogonal transformation such as discrete cosine transform where redundancy in a spatial direction is used.

[0065] In 2003, through the cooperative work of the Joint Technical Committee (ISO/IEC) of the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC) and the International Telecommunication Union Telecommunication Standardization Sector (ITU-T), a coding system called AVC (a default number of 14496-10 is assigned in ISO/IEC and a default number of H.264 is assigned in ITU-I) has been established as a global standard. In AVC, a median value of the respective motion vectors of a plurality of adjacent blocks of a target block is basically defined to be a vector prediction. If the size of a prediction block is not square and a reference index of a specific adjacent block of a target block is the same as a reference index of the target block, a motion vector of the specific adjacent block is defined to be a vector forecast.

[0066] Currently, through the cooperative work of the Joint Technical Committee (ISO/IEC) of the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC) and the International Telecommunication Union Telecommunication Standardization Sector (ITU-T), standardization of an encoding system called HEVC is under consideration.

[0067] In HEVC standardization, a confluence mode is under consideration in which a single candidate block is selected from a group of candidate blocks composed of candidate blocks, which are a plurality of adjacent blocks of a block and target blocks of another representation which is decoded such that the selected candidate block information is encoded and decoded and such that the selected candidate block motion information is used as the target block motion information. A vector prediction mode is under consideration in which a single candidate block is selected from a group of candidate blocks composed of candidate blocks, which are a plurality of adjacent blocks of the block and target blocks of another representation that are decoded, such that information of the selected candidate block is encoded and decoded and such that a motion vector of the selected candidate block is used as a vector prediction of the target block.

FIRST MODALITY CODING BLOCK

[0068] In the present embodiment, a representation signal that has been inserted is partitioned into maximum coding block units, and maximum coding blocks that have been partitioned are processed in scan-by-scan order. A coding block has a hierarchical structure, and smaller coding blocks can be obtained by quartering the coding block sequentially in consideration of coding efficiency and the like. The quartered cipher blocks are encoded in zigzag scan order. The minimal coding blocks that cannot be made smaller are referred to as minimal coding blocks. Coding blocks represent coding units. If the number of partition occurrences is zero in a maximum code block, the maximum code block also represents a code block. In the present embodiment, a maximum coding block represents 64 pixels x 64 pixels and a minimum coding block represents 8 pixels x 8 pixels.

[0069] Figures 1A and 1B are diagrams to explain the coding blocks. In an example shown in Figure 1A, a cipher block is partitioned into 10 chunks. CU0, CU1, and CU9 represent the 32 pixel x 32 pixel code blocks, CU2, CU3, and CU8 represent the 16 pixel x 16 pixel code blocks, and CU4, CU5, CU6, and CU7 represent the 8 pixel x code blocks 8 pixels. In an example shown in Figure 1B, a block of code is partitioned into a chunk.

FORECAST BLOCK

[0070] In the present embodiment, a coding block is further partitioned into prediction blocks (also referred to as partitions). A coding block is partitioned into greater than or equal to a prediction block according to a prediction block size type (also referred to as partitioning type or partition type). Figures 2A through 2D are diagrams to explain the prediction block size types. Figure 2A shows 2Nx2N where a cipher block is not partitioned. Figure 2B shows 2NxN where a cipher block is split in two in a horizontal direction. Figure 2C shows Nx2N where a cipher block is split in two in a vertical direction. Figure 2D shows NxN where a cipher block is quartered in a horizontal and a vertical direction. 2Nx2N consist of a single prediction block 0. Both 2NxN and Nx2N each consist of two prediction blocks: a prediction block 0 and a prediction block 1. NxN consist of four prediction blocks: a prediction block 0, a prediction block 1, a prediction block 2, and a prediction block 3. Encoding is performed in the order of a prediction block 0, a prediction block 1, a prediction block 2, and a prediction block 3.

[0071] Figure 3 is a diagram for explaining the prediction block sizes according to the number of partition occurrences of a coding block and types of prediction block size. For the prediction block sizes in the present embodiment, there are 13 prediction block sizes starting from 64 pixels x 64 pixels, where the number of CU partition occurrences is 0 and a type of prediction block size is 2Nx2N , at 4 pixels x 4 pixels, where the number of CU partition occurrences is 3 and a prediction block size type is NxN. For example, a block of code can be split into two in a horizontal direction or a vertical direction in an asymmetrical manner.

[0072] In the present embodiment, a maximum coding block represents 64 pixels x 64 pixels and a minimum coding block represents 8 pixels x 8 pixels. However, the maximum cipher block and minimum cipher block are not limited to this combination. Partition patterns of a prediction block are shown as those in Figures 2A through 2D. However, the partition patterns are not limited to this as long as the partition patterns are a combination of the patterns where a prediction block is partitioned into greater than or equal to one chunk.

REPRESENTATIONS AND PORTIONS

[0073] Representations and portions are general concepts that are used in AVC and the like. Therefore, their explanations are omitted. Also, since I-representations, P-representations, B-representations, I-portions, P-portions, and B-portions are also general concepts, explanations of them are omitted. Henceforth, an image may be referred to as a representation.

PREDICTION CODING MODE

[0074] In the present embodiment, the motion compensation prediction and the amount of coding vectors can be changed for each prediction block. An explanation is now given using Figure 4 with respect to an example of a predictive coding mode with which the motion compensation prediction and the number of coding vectors are associated. Figure 4 is a diagram to explain the predictive coding modes.

[0075] The prediction coding modes shown in Figure 4 include PredL0 where a prediction direction of the motion compensation prediction is uni-prediction (prediction L0) and the amount of coding vectors is 1, PredL1 where a direction of motion compensation prediction prediction is the uni-prediction (L1 prediction) and the amount of coding vectors is 1, PredBI where a motion compensation prediction prediction direction is the dual prediction (BI prediction) and the amount of coding vectors is 2 and a confluence mode (CONFLUENCE) where a prediction direction of motion compensation prediction is uni-prediction (L0 prediction/L1 prediction) or dual prediction (BI prediction) and the number of vectors coding mode is 0. There is also an intra mode (Intra), which is a prediction coding mode in which motion compensation prediction is not performed. In these modes, PredL0, PredL1, and PredBI are the vector prediction modes.

[0076] In confluence mode, a forecast direction can be any of the L0 forecast, the L1 forecast and the BI forecast. This is due to the fact that the prediction direction of a candidate block selected from a group of candidate blocks is passed without any change as the confluence mode prediction direction or the confluence mode prediction direction is derived from from the decoded information. Also, an encoding vector is not encoded in confluence mode. This is due to the fact that a motion vector of a candidate block selected from a group of candidate blocks is passed without any change as a confluence mode coding vector or the confluence mode coding vector is derived through a predetermined rule.

REFERENCE INDEX

[0077] The present embodiment allows an ideal reference representation to be selected from a plurality of reference representations in the motion compensation prediction for improving the accuracy of the motion compensation prediction. Therefore, a reference representation used in motion compensation prediction is encoded together with a coding vector as a reference representation index. A reference representation index used in motion compensation prediction has a numerical value greater than or equal to 0. Reference type indexes include an L0 reference index (also referred to as an L0 prediction reference index) and a L1 benchmark (also referred to as an L1 forecast benchmark). If the motion compensation forecast is uni-forecast, either the L0 benchmark and the L1 benchmark will be used as a benchmark. If the motion compensation forecast is dual forecast, two benchmarks, the L0 benchmark and the L1 benchmark, will be used (Figure 4). As a reference index bitstream, a Truncated Unary bitstream described later is used.

[0078] A reference index is not encoded in confluence mode. This is due to the fact that a reference index of a candidate block selected from a group of candidate blocks is passed without any change as a confluence mode reference index or the confluence mode reference index is derived through a predetermined rule.

REFERENCE REPRESENTATION LIST

[0079] In the present embodiment, greater than or equal to a reference representation that can be used in motion compensation prediction is added to a reference representation list before and, indicating a reference representation added to the representation list reference using a reference index, the reference representation is determined and used in the motion compensation forecast. Types of reference representation lists include an L0 reference representation list (also referred to as an L0 forecast reference representation list) and an L1 reference representation list (also referred to as a forecast reference representation list L1). If the motion compensation forecast is uni-forecast, both the L0 forecast, where a reference representation in the L0 reference representation list is used, and the L1 forecast, where a reference representation in the L0 reference representation list is used. L1 reference is used, will be used. If the motion compensation forecast is the dual forecast, the BI forecast, in which both the L0 reference representation list and the L1 reference representation list are used, will be used. It is observed that a reference index L0 indicates a reference representation in the reference representation list L0 and that a reference index L1 indicates a reference representation in the reference representation list L1. CONFLUENCE INDEX

[0080] In the case of a confluence mode in the present embodiment, using, as a group of candidate blocks, a plurality of adjacent blocks in a representation and in target blocks in and around a prediction block in the same position situated in the same position as a target prediction block in another representation that is coded, where a candidate block having an optimal prediction coding mode, a motion vector and a reference index is selected from the group of candidate blocks of in order to encode and decode a confluence index to indicate the selected candidate block. A confluence index is only used during confluence mode (Figure 4). In this case, the maximum number of confluence indices (also referred to as the maximum number of confluence candidates) is specified via a portion header. The maximum amount of confluence candidates will be described later. If the maximum number of confluence candidates is 5, the confluence index is an integer from 0 to 4.

[0081] Hereinafter, move information of a candidate block submitted to a confluence index is referred to as a confluent move information candidate, and a collection of confluent move information candidates is referred to as a move information candidate list confluent. Henceforth, motion information includes a prediction direction, a motion vector, and a reference index.

[0082] An explanation will now be given regarding the relationships between confluence indices and bit streams. Figure 5 is a diagram for explaining the relationships between confluence indices and bitstreams when the maximum number of confluence candidates is 5. As a bitstream of a confluence index, a Unary bitstream is used. Truncated. If a confluence index is 0, a bitstream will be "0". If a confluence index is 1, a bitstream is "10". If a confluence index is 2, a bitstream is "110". If a confluence index is 3, a bitstream is "1110". If a confluence index is 4, a bitstream is "1111". In this way, the confluence indices and bitstreams are defined such that the respective bitstreams become shorter as the confluence indices become smaller. Therefore, by assigning a small confluence index to a candidate block with high selectivity, the coding efficiency can be improved.

[0083] If the maximum number of confluence candidates is 4, a bitstream is "0" if a confluence index is 0, a bitstream is "10" if a confluence index is 1, a bitstream will be "110" if a confluence index is 2, and a bitstream will be "111" if a confluence index is 3. If maximum number of confluence candidates is 3, a bitstream will be "0" if an index of confluence is 0, a bitstream is "10" if a confluence index is 1, and a bitstream is "11" if a confluence index is 2. If the maximum number of confluence candidates is 2, a confluence bits will be "0" if a confluence index is 0 and a bitstream will be "1" if a confluence index is 1. If the maximum number of confluence candidates is 1, a confluence index of 0 will not be encoded in a stream of bits and the confluence index will be implicitly rendered as 0 at decoding time. O.

[0084] An explanation is now given with respect to the relationships between a candidate list of confluent movement information and the confluence indices. A confluence index of 0 represents a first (0th) confluent motion information candidate in a list of confluent motion information candidates. Henceforth, a confluence index m represents an mth confluent movement information candidate in the confluent movement information candidate list, where m is an integer from 0 to [(maximum number of confluence candidates) - 1] .

VECTOR PREDICTION INDEX

[0085] In order to improve the accuracy of a vector prediction in the present embodiment, using, as a group of candidate blocks, a plurality of adjacent blocks in a representation and target blocks in and around a prediction block in the same position situated at the same position as a target prediction block in another representation that is coded, wherein a candidate block has an ideal motion vector as a prediction vector is selected from the group of candidate blocks in order to code and decode a vector prediction index to indicate the selected candidate block. If the motion compensation prediction is uni-prediction, a vector prediction index is used. If the motion compensation prediction is dual prediction, two vector prediction indices are used (Figure 4). The maximum number of vector prediction indices (also referred to as the maximum number of vector prediction candidates) is 2, and a vector prediction index is an integer of either 0 or 1. The maximum number of vector prediction indices is defined to be 2 in that case. However, the maximum amount of vector prediction indices is not limited to this as long as the maximum amount of vector prediction indices is greater than or equal to 2. An explanation is now given regarding the relationships between the vector prediction indices. vector and bit streams. As a bitstream of a vector prediction index, a Truncated Unary bitstream is used. A bitstream of a vector prediction index is "0" when the vector prediction index is 0, and a bitstream of a vector prediction index is "1" when the vector prediction index is 1.

[0086] Hereinafter, a motion vector of a candidate block submitted to a vector prediction index is referred to as a vector prediction candidate and a collection of vector prediction candidates is referred to as a vector prediction candidate list . A vector prediction index of 0 represents a first (0th) vector prediction candidate in the list of vector prediction candidates. Henceforth, a vector prediction index m represents an mth vector prediction candidate in the list of vector prediction candidates, where m is an integer from 0 to [(maximum number of vector prediction candidates) - 1].

POC

[0087] In some embodiments of the present invention, the POC (Picture Order Count) is used as temporal information (distance information) of a reproduction. The POC is a counter that indicates the display order of a representation and is the same as the one defined in AVC. In this case, it is assumed that a POC increases by one when the display order of a representation increases by one. Therefore, a temporal difference (distance) between the representations can be acquired based on a POC difference between the representations.

SYNTAX

[0088] An explanation is given with respect to a syntax example of a prediction block according to the present embodiment. Figure 6 is a diagram explaining the syntax according to the present embodiment. Figure 6 shows an example of a syntax structure of a PPS (Picture Parameter Set), a slice header (Slice Header), a coding tree (Coding Tree), a coding block ( Coding Block) and a prediction block (Prediction Unit).

[0089] A PPS is a set of parameters that defines a group of parameters to determine the properties of a representation. For a PPS, a predetermined maximum value of an L0 prediction benchmark (num_ref_idx_l0_default_active_minus1), a predetermined maximum value of an L1 prediction benchmark (num_ref_idx_l1_default_active_minus1), and a temporal candidate use enable flag (enable_temporal_mvp_flag) that can to be used in the representation are defined.

[0090] A temporal candidate use permission flag is a one-bit code that takes the value of 0 or 1 and is a flag to restrict the use of a motion vector and a reference index in a ColPic. If the temporal candidate use permission flag is 1, the motion vector and reference index in ColPic become available. If the temporal candidate use permit flag is 0, the motion vector and reference index in ColPic become unavailable. In the embodiment, an explanation is made by assuming that the temporal candidate use permission flag is 1.

[0091] A portion header is a header that defines a group of parameters to determine the properties of a portion. In the portion header, a flag (num_ref_idx_active_override_flag) to change the maximum value of a reference index that can be used on the slice when the slice is a P-slice or a B-slice, the maximum value of an L0 prediction reference index ( num_ref_idx_l0_active_minus1) and the maximum value of an L1 prediction reference index (num_ref_idx_l1_active_minus1) are defined.

[0092] If the flag to change the maximum value of the reference index is 1, the maximum value of the prediction reference index L0 (num_ref_idx_l0_active_minus1) that can be used in the slice is defined in the portion header and if the slice is a slice B, the maximum value of L1 prediction reference index (num_ref_idx_l1_active_minus1) that can be used in the slice will be further defined in the portion header and used.

[0093] If the flag for changing the maximum value of the reference index is 0, the predetermined maximum value of the prediction reference index L0 that is defined in the PPS will be used as the maximum value of the prediction reference index L0 that can be used in the slice and the predetermined maximum value of L1 prediction benchmark that is set in the PPS will be applied as the maximum value of L1 prediction benchmark that can be used in the slice.

[0094] If the slice is a P slice or a B slice, 5_minus_max_num_merge_cands, which is a parameter to determine the maximum amount of confluence candidates, will be set. 5_minus_max_num_merge_cands is a parameter to determine the maximum number of merging candidates and is an integer from 0 to 4. The maximum number of merging candidates (MaxNumMergeCand) is calculated by subtracting 5_minus_max_num_merge_cands from 5. Truncated Unary Bitstream is used for a bitstream of 5_minus_max_num_merge_cands. In the present embodiment, an explanation is given by assuming that 5_minus_max_num_merge_cands is 0 and that the maximum number of confluence candidates is 5. In the coding tree, coding block partition information is managed. In the coding tree, split_coding_unit_flag is set. If the split_coding_unit_flag is 1, the coding tree will be partitioned into four coding trees. If the split_coding_unit_flag is 0, the coding tree represents one coding block.

[0095] In the coding block, a skip mode flag (skip_flag), a preview mode (pred_mode) and a preview block size type (part_mode) are defined. The coding block is partitioned into one, two or four prediction blocks according to the signaling hop mode and type of prediction block size.

[0096] Prediction mode shows whether the coding block is a coding block in which intraprediction is performed or a coding block in which interprediction (motion compensation prediction) is performed. If the flag jump mode is 1, a jump mode is implemented which has a prediction block in the jump mode. The number of partition occurrences of a coding block is also referred to as a coding block depth (coding tree).

[0097] A confluence flag (merge_flag), a confluence index (merge_idx), an interforecast type (inter_pred_type), a forecast reference index L0 (ref_idx_l0), a forecast vector difference are defined in the forecast block L0 (mvd_l0[0], mvd_l0[1]), a prediction vector index from the forecast L0 (mvp_idx_l0), a reference index from the forecast L1 (ref_idx_l1), a vector difference from the forecast L1 (mvd_l1[0], mvd_l1[1]) and a forecast vector index of the L1 forecast (mvp_idx_l1). In vector difference, [0] represents the horizontal component and [1] represents the vertical component.

[0098] In this case, inter_pred_type shows a motion compensation forecast forecast direction (also referred to as an interforecast type) and includes three types: Pred_L0 (uni-forecast of L0 forecast); Pred_L1 (uni-forecast of L1 forecast); and Pred_BI (bi forecast dual forecast). If inter_pred_type is Pred_L0 or Pred_BI, information related to L0 prediction is defined. If inter_pred_type is Pred_L1 or Pred_BI, L1 prediction related information is set. In a slice P, inter_pred_type uniquely represents Pred_L0. Therefore, inter_pred_type is omitted.

[0099] In case of jump mode, a prediction block is a coding block in which interprediction is performed and a confluence mode is used as a prediction coding mode. Therefore, a confluence index is defined in the case of jump mode.

[00100] The syntax according to the present embodiment is defined as shown in Figure 6. However, the syntax is not limited to this since the coding blocks and prediction blocks have a plurality of block sizes, a representation of reference is used and both confluence mode and vector prediction mode can be used.

[00101] An explanation is given hereinafter, together with the Figures, regarding the details of a motion representation encoding device, a motion representation encoding method and a motion representation encoding program and the decoding device representation, a motion representation decoding method and a motion representation decoding program according to a preferred embodiment of the present invention. In explanations of Figures, the same elements may be denoted by the same reference numerals and duplicate explanations will be omitted.

CONFIGURATION OF THE ENCRYPTION DEVICE REPRESENTATION IN MOTION 100

[00102] Figure 7 shows the configuration of a moving representation coding device 100 according to a first embodiment. The motion representation encoding device 100 is a device which encodes a motion representation signal into the prediction block units to perform motion compensation prediction. Determining a slice type, the maximum value of a reference index that can be used in a slice, partitioning a coding block, determining a jump mode, determining a type of block size prediction, determining a prediction block size and a position within a coding block of a prediction block (also referred to as the position information or a prediction block number of a prediction block), and determining whether a for intra prediction coding mode will be determined by a coding control unit 112 outside the motion representation coding device 100 and fed to the motion representation coding device 100. A reference representation list is constructed through a reference representation list building unit 113 out of the moving representation encoding device 100 and supplied to the reference representation encoding device moving representation 100. An explanation is given regarding a case in which a predictive encoding mode is not intra. An explanation is given regarding a representation B (slice B), which corresponds to the double prediction, in the first embodiment unless otherwise specified. For a representation P (slice P), which does not correspond to the double prediction, the prediction L1 needs to be omitted.

[00103] The moving representation coding device 100 is achieved through hardware such as, for example, an information processing device or the like provided with a CPU (Central Processing Unit), a frame memory, a hard disk and similar. Through operation of the above constituted elements, the moving representation encoding device 100 achieves the functional constituted elements explained below. The slice type, the reference representation list, the maximum value of the reference index that can be used in the slice and the position information, the prediction block size and the prediction direction of the motion compensation prediction of a target prediction blocks are assumed to be shared in the motion representation encoding device 100 and are therefore not displayed.

[00104] The moving representation coding device 100 according to the first embodiment includes a prediction block representation derivation unit 101, a subtraction unit 102, a prediction error coding unit 103, a bitstream generation 104, a prediction error decoding unit 105, a motion compensation unit 106, an addition unit 107, a motion vector detection unit 108, a motion information generation unit 109 , a frame memory 110 and a motion information memory 111.

FUNCTION AND OPERATION OF THE MOTION REPRESENTATION CODING DEVICE 100

[00105] An explanation is given below regarding the function and operation of each component. The prediction block representation derivation unit 101 derives a representation signal of a target prediction block from a representation signal supplied from a terminal 10 based on the position information and the prediction block size of the target. prediction block and supplies the prediction block representation signal to the subtraction unit 102, the motion vector detection unit 108 and the motion information generation unit 109.

[00106] The motion vector detection unit 108 detects the respective motion vectors and the respective reference indices that show the reference representations for the prediction L0 and the prediction L1 in the representation signal supplied by the representation derivation unit of prediction block 101 and respective representation signals corresponding to a plurality of reference representations stored therein. The motion vector detection unit 108 supplies the respective motion vectors of prediction L0 and prediction L1 and the respective reference indices of prediction L0 and prediction L1 to motion information generation unit 109. motion vector detection unit 108 uses the respective representation signals corresponding to the plurality of reference representations stored therein as reference representations, the motion vector detection unit 108 can also use the reference representations stored in the memory of frame 110.

[00107] In a commonly practiced method of detecting a motion vector, an evaluation value of an error between a representation signal of a target representation and a prediction signal of a reference representation moved from the same position through a predetermined amount of displacement is calculated and an amount of displacement that results in the smallest error evaluation value is defined to represent a motion vector. If there are a plurality of reference representations, a motion vector will be detected for each of the reference representations and a reference representation with the smallest error evaluation value will be selected. Like the error evaluation value, a SAD (Sum of Absolute Difference) that shows the sum of an absolute difference, an MSE (Mean Square Error) that shows a mean error square or similar can be used. It is also possible to add a motion vector encoding amount to the error evaluation value in order to perform the evaluation.

[00108] The motion information generation unit 109 determines a prediction coding mode based on the respective motion vectors of the prediction L0 and the prediction L1 and the respective reference indices of the prediction L0 and the prediction L1 supplied through the unit of motion vector detection 108, a group of candidate blocks supplied via the motion information memory 111, the reference representations stored in the frame memory 110 that are indicated via the reference indices, and the representation signal supplied by the motion vector unit 110. prediction block representation derivation 101.

[00109] Based on the prediction coding mode that has been determined, the movement information generation unit 109 supplies the bit stream generation unit 104 with a confluence signaling, a confluence index, a prediction prediction direction of motion compensation, the respective reference key figures of the L0 forecast and the L1 forecast, the respective vector differences of the L0 forecast and the L1 forecast, and the respective vector forecast key figures of the L0 forecast and the L1 forecast, as required. The motion information generating unit 109 supplies the motion compensation unit 106 and the motion information memory 111 with the forecast direction of the motion compensation forecast, the respective reference indices of forecast L0 and forecast L1, and the respective motion vectors of forecast L0 and forecast L1. Details of motion information generation unit 109 will be described later.

[00110] If the prediction direction of the motion compensation prediction supplied by the motion information generation unit 109 is the LN prediction, the motion compensation unit 106 will perform motion compensation on a reference representation in the frame memory 110 which is indicated by a prediction reference index LN supplied by the motion information generation unit 109 on the basis of a prediction motion vector LN supplied by the motion information generation unit 109 so as to generate a prediction signal for the LN forecast. N is 0 or 1. If the forecast direction of motion compensation forecast is dual forecast, an average value of the respective forecast signals for forecast L0 and forecast L1 is set to represent the forecast signal. The respective forecast signals for the L0 forecast and the L1 forecast can be weighted. The motion compensation unit 106 supplies the prediction signals to the subtraction unit 102.

[00111] The subtraction unit 102 subtracts the representation signal supplied by the prediction block representation derivation unit 101 and the prediction signals supplied by the motion compensation unit 106 in order to calculate a prediction error signal and supply the prediction error signal to the prediction error coding unit 103.

[00112] The prediction error coding unit 103 generates the prediction error coding data by performing a process such as, for example, orthogonal transformation, quantization or the like on the prediction error signal provided from the subtraction unit 102 and supplies the prediction error coding data to the bitstream generation unit 104 and the prediction error decoding unit 105.

[00113] The bit stream generation unit 104 submits the prediction error coding data supplied by the prediction error coding unit 103 and the confluence signaling, the confluence index, the prediction direction (type of interprediction ) motion compensation forecast, the respective reference key figures of the L0 forecast and the L1 forecast, the respective vector differences of the L0 forecast and the L1 forecast, and the respective vector forecast key figures of the L0 forecast and the L1 forecast supplied by Motion information generation unit 109 for entropy encoding according to the syntax order shown in Fig. 6 so as to generate a bitstream and supply the bitstream to terminal 11 as a bitstream. Entropy coding is performed using a method including variable length coding such as arithmetic coding, Huffman coding or the like.

[00114] The bitstream generation unit 104 multiplexes the partition information for the coding block, the type of prediction block size, the position in the coding block of the prediction block and the used prediction coding mode in the motion representation coding device 100 in the bit stream together with a Sequence Parameter Set (SPS) that defines a group of parameters for determining the properties of the bit stream, a Sequence Parameter Set (PPS) (Picture Parameter Set) that defines a group of parameters for determining the properties of the representation, where a portion header defines a group of parameters for determining the properties of the slice and the like.

[00115] The prediction error decoding unit 105 generates a prediction error signal by performing a process such as, for example, inverse quantization, inverse orthogonal transformation or the like on the prediction error coding data supplied by the coding unit the prediction error signal 103 and supplies the prediction error signal to the summing unit 107. The summing unit 107 adds the prediction error signal supplied by the prediction error decoding unit 105 and the prediction signals supplied by the prediction motion compensation 106 so as to generate a decoding representation signal and supply the decoding representation signal to frame memory 110.

[00116] The frame memory 110 stores the decoding representation signal supplied by the addition unit 107. For a decoded representation in which the decoding of the entire representation has been completed, the frame memory 110 stores, together with the POC of the representation reference, a predetermined quantity greater than or equal to a representation of the same as a reference representation. The frame memory 110 supplies a stored reference representation signal to the motion compensation unit 106 and the motion information generation unit 109. A storage area that stores the reference pictures is controlled by a FIFO (First First In, First Out). In that case, the POC of the reference representation is stored in the frame memory 110. However, as long as the reference representation and the POC can be uniquely identified, the storage of the POC of the reference representation is not limited to that. Also, the POC is not shown in the Figure as being shared in the moving representation coding device 100, the coding control unit 112 and the reference representation list building unit 113.

[00117] The motion information memory 111 defines the motion information of a block adjacent to the target prediction block to represent a group of special candidate blocks.

[00118] Also, the motion information memory 111 defines the motion information in a prediction block at the same position, which is situated at the same position as the target prediction block, in a ColPic and a block around the block of prediction at the same position to represent a group of temporal candidate blocks. The motion information memory 111 supplies the special candidate block group and the temporal candidate block group to the motion information generation unit 109 as the candidate block groups. Motion information memory 111 is synchronized with frame memory 110 and is controlled by the FIFO (First In First Out) method.

[00119] A ColPic is a representation that has been decoded and that is different from a representation with a target prediction block and is stored in the frame memory 110 as a reference representation. In the first embodiment, a ColPic is a reference representation that is decoded immediately before a target representation. In the first embodiment, the ColPic is a reference representation that is decoded immediately before a target representation. However, since the ColPic is a decoded representation, the ColPic can be, for example, a reference representation immediately before or immediately after the target representation in the display order or a 0th reference representation from the reference representation list of L0 prediction or L1 prediction and can be specified in a bit stream.

[00120] An explanation is now given regarding a method of managing the movement information in the movement information memory 111. The movement information is stored in the units of the minimum prediction blocks in each memory area. Each memory area stores at least one forecast direction, a forecast motion vector L0, a forecast reference index L0, a forecast motion vector L1, and a forecast reference index L1.

[00121] If the prediction coding mode is intra-mode, (0,0) will be stored as the respective motion vectors of prediction L0 and prediction L1 and "-1" will be stored as the respective reference indices of forecast L0 and forecast L1. Henceforth, in (H,V) of a motion vector, H represents a horizontal component and V represents a vertical component. As long as it can be determined that a mode is used in which motion compensation prediction is not performed, the "-1" value of the reference indices can be any value. Hereinafter, what is simply expressed as "block" represents the minimum prediction block unit unless otherwise specified. The same holds for a block outside the area. As in the intra-mode case, (0,0) is stored as the respective L0 forecast and L1 forecast motion vectors and "-1" is stored as the respective L0 forecast and L1 forecast reference indices. An LX direction (X is 0 or 1) is a valid means that a benchmark in the LX direction is greater than or equal to 0. The LX direction is an invalid (not valid) means that the benchmark in the direction LX is "-1".

REFERENCE REPRESENTATION LIST BUILDING UNIT OPERATION 113

[00122] Next, an explanation is given regarding the operation of the reference representation list building unit 113. The reference representation list building unit 113 builds a reference representation list according to the type of slice supplied from the coding control unit 112. If the slice type is a P-slice or a B-slice, the reference representation list building unit 113 will build a reference representation list L0. If slice type is B slice, reference representation list building unit 113 will build reference representation list L1.

[00123] Figure 8 is a flowchart that explains the construction operation of a L0 reference representation list. With the use of Fig. 8, an explanation is given regarding the L0 reference representation list construction operation. Firstly, the reference representation list building unit 113 arranges the reference representations having a POC less than the POC of the target representation in POC descending order up to the quantity of reference representations available in the reference representation list L0 (RefPicListL0) (S400). Subsequently, the reference representation list building unit 113 arranges the reference representations having a POC greater than the POC of the target representation in POC ascending order up to the quantity of reference representations available in the reference representation list L0 ( RefPicListL0) (S401). In this case, the amount of reference representations available in the list of reference representation L0 (RefPicListL0) is a value obtained by adding 1 to the maximum value of the prediction reference index L0 (num_ref_idx_l0_active_minus1).

[00124] Figure 9 is a flowchart explaining the operation of building an L1 reference representation list. With the use of Fig. 9, an explanation is given regarding L1 reference representation list construction operation. Firstly, the reference representation list building unit 113 arranges the reference representations having a POC greater than the POC of the target representation in POC ascending order up to the amount of available reference representations in the reference representation list L1 ( RefPicListL1) (S410). Subsequently, the reference representation list building unit 113 arranges the reference representations having a POC smaller than the POC of the target representation in POC descending order up to the quantity of reference representations available in the reference representation list L1 (RefPicListL1) (S411). In this case, the amount of reference representations available in the reference representation list L1 (RefPicListL1) is a value obtained by adding 1 to the maximum value of the prediction reference index L1 (num_ref_idx_l1_active_minus1).

[00125] As described above, the reference representations are arranged in the reference representation list L0 starting from a reference representation which is displayed earlier than the target representation and which is closer to the target representation. Reference representations are arranged in reference representation list L1 starting from a reference representation which is displayed after the target representation and which is closest to the target representation.

[00126] Also, if the quantity of reference representations available in the reference representation list L0 (RefPicListL0) is greater than the quantity of reference representations displayed earlier than the target representation, the reference representation list L0 will have the same reference representations than those in reference representation list L1. If the quantity of reference representations available in the reference representation list L1 (RefPicListL1) is greater than the quantity of reference representations displayed after the target representation, the reference representation list L1 will have the same reference representations as those in the list of reference representation L0. This allows for dual prediction using the same reference representations.

CONFIGURATION OF THE MOVEMENT INFORMATION GENERATION UNIT 109

[00127] An explanation is now given regarding the detailed configuration of the motion information generation unit 109. Figure 10 shows the configuration of the motion information generation unit 109. The motion information generation unit 109 includes a a vector prediction mode determining unit 120, a confluence mode determining unit 121, and a predictive coding mode determining unit 122. A terminal 12, a terminal 13, a terminal 14, a terminal 15, a terminal 16, a terminal 50 and a terminal 51 are connected to the motion information memory 111, the motion vector detection unit 108, the frame memory 110, the prediction block representation derivation unit 101, the generating bit stream 104, motion compensation unit 106 and motion information memory 111, respectively.

FUNCTION AND OPERATION OF THE MOVEMENT INFORMATION GENERATION UNIT 109

[00128] An explanation is given below regarding the function and operation of each component. The vector prediction mode determining unit 120 determines an interprediction type based on a group of candidate blocks supplied by the terminal 12, the respective L0 prediction and L1 prediction motion vectors, and the respective L0 prediction reference indices and forecast L1 supplied by terminal 13, a reference representation indicated by a reference index supplied by terminal 14, and a representation signal supplied by terminal 15. Based on the interforecast type, the vector forecast mode determination unit 120 selects the respective vector prediction indices of the L0 prediction and the L1 prediction so as to calculate the respective vector differences of the L0 prediction and the L1 prediction and calculate a prediction error and also calculate a rate distortion evaluation value. The vector forecast mode determining unit 120 then supplies the motion information, vector differences, vector forecast indices and rate distortion assessment value based on the type of interforecast to the vector forecast mode determination unit. predictive coding mode determination 122. Details of the vector predictive mode determination unit 120 will be described later.

[00129] The confluence mode determination unit 121 builds a candidate list of confluent movement information from the group of candidate blocks supplied by terminal 12, the reference representation supplied by terminal 14 and the representation signal supplied by terminal 15, select a confluent motion information candidate from the list of confluent motion information candidates so as to determine a confluence index and calculate a rate distortion assessment value. The confluence mode determination unit 121 then supplies the candidate movement information of confluent movement information, the confluence index, and the rate distortion assessment value to the prediction coding mode determination unit 122 The details of the confluence mode determining unit 121 will be described later.

[00130] The prediction coding mode determination unit 122 determines a confluence signal by comparing the rate distortion evaluation value supplied by the vector prediction mode determination unit 120 and the vector distortion evaluation value rate supplied by the confluence mode determination unit 121.

[00131] If the rate distortion assessment value for the vector prediction mode is less than the rate distortion assessment value for the confluence mode, the prediction coding mode determination unit 122 sets the flag from confluence to "0". The prediction coding mode determining unit 122 supplies to the terminal 16 the confluence signaling and the interprediction type, the reference indices, the vector differences and the vector prediction index supplied by the mode determining unit vector prediction mode determination unit 120 and supplies the motion information supplied by the vector prediction mode determining unit 120 to terminals 50 and 51.

[00132] If the rate distortion evaluation value for the confluence mode is the rate distortion evaluation value for the vector prediction mode or less, the prediction coding mode determination unit 122 sets the flag from confluence to "1". The prediction coding mode determining unit 122 supplies the terminal 16 with the confluence signaling and the confluence index supplied by the confluence mode determining unit 121 and supplies the movement information supplied by the confluence mode determining unit confluence 121 to terminals 50 and 51. A specific method of calculating a rate distortion assessment value is not the main point of the present invention and a detailed explanation thereof is therefore omitted. From a prediction error and a coding quantity, a prediction error quantity by a coding quantity is calculated. The rate skew assessment value is an assessment value that has a property that the coding efficiency becomes higher as the rate skew assessment value becomes smaller. Therefore, by selecting a predictive coding mode with a small rate distortion evaluation value, the coding efficiency can be improved.

CONFLUENCE MODE DETERMINATION UNIT CONFIGURATION 121

[00133] An explanation is now given regarding the detailed configuration of the confluence mode determination unit 121. Figure 11 is a diagram explaining the configuration of the confluence mode determination unit 121. The confluence mode determination unit The confluence 121 includes a confluent move information candidate list construction unit 140 and a confluent move information candidate list construction unit 141. A confluent move information candidate list construction unit 140 is likewise provided at a motion representation decoding device 200 which decodes a bit stream generated by the motion representation encoding device 100 according to the first embodiment, and an identical confluent motion information list is constructed each in the coding device representation device 100 and the motion representation decoding device move 200.

FUNCTION AND OPERATION OF THE CONFLUENCE MODE DETERMINATION UNIT 121

[00134] An explanation is given below regarding the function and operation of each component. Figure 12 is a flowchart explaining the operation of the confluence mode determining unit 121. The confluent motion information candidate list construction unit 140 builds a confluent motion information candidate list that includes the information candidates. of the maximum number of confluence candidates from the candidate block group supplied via terminal 12 (S100) and supplies the candidate list of confluent movement information to the confluent movement information selection unit 141. The configuration Details of the confluent movement information candidate list construction unit 140 will be described later.

[00135] The confluent movement information selection unit 141 selects an ideal confluent movement information candidate from the confluent movement information candidate list supplied from the confluent movement information candidate list construction unit 140 (S101), determines a confluence index that serves as the information indicating the selected confluent movement information candidate (S102), and supplies the confluence index to terminal 17.

[00136] An explanation is now given with respect to an ideal confluent motion information candidate selection method. A prediction error amount is calculated from the reference representation obtained by performing the motion compensation prediction based on the prediction direction, the motion vector, and the reference index of the confluent motion information candidate and supplied through from terminal 14 and from the representation signal supplied by terminal 15. A rate distortion evaluation value is calculated from the amount of confluence index encoding and the amount of prediction error and a candidate of confluent motion information with the lowest rate distortion assessment value is selected with the ideal confluent motion information candidate.

CONFLUENT MOVEMENT INFORMATION CANDIDATE LIST BUILDING UNIT CONFIGURATION 140

[00137] An explanation of detailed configuration of the confluent movement information candidate list construction unit 140 is now given. Figure 13 is a diagram for explaining the configuration of the confluent movement information candidate list construction unit 140 A terminal 19 is connected to the confluent motion information candidate selection unit 141. The confluent motion information candidate list building unit 140 includes a spatial confluent motion information candidate derivation unit 160, a derivation unit confluent motion information candidate list building unit 161, a confluent motion information candidate list building unit 162, a redundant confluent motion information candidate deletion unit 163, a first motion information candidate supply unit confluent 164 and a second candid supply unit confluent movement information act 165.

FUNCTION AND OPERATION OF THE CONFLUENT MOVEMENT INFORMATION CANDIDATE LIST BUILDING UNIT 140

[00138] An explanation is given below regarding the function and operation of each component. Figure 14 is a flowchart for explaining the operation of the confluent motion information candidate list building unit 140.

[00139] First, the spatial confluent motion information candidate derivation unit 160 derives the spatial confluent motion information candidates from zero to the maximum amount of spatial confluent motion information candidates from the candidate block group 12 (S110) and supply the spatial confluent motion information candidates to the confluent motion information candidate list building unit 162. The detailed operation of the spatial confluent motion information candidate derivation unit 160 will be described later. Also, descriptions will be made later regarding the maximum amount of spatial confluent motion information candidates.

[00140] Next, the confluent movement information candidate list construction unit 140 checks whether the temporal candidate use permission flag (enable_temporal_mvp_flag) is 1 (S111).

[00141] If the temporal candidate use permission flag is 1 (S at S111), the temporal confluent movement information candidate derivation unit 161 will derive the temporal confluent movement information candidates from zero to the maximum amount of temporal confluent motion information candidates from the group of temporal candidate blocks supplied by terminal 12 (S112) and will feed the temporal confluent motion information candidates to the confluent motion information candidate list building unit 162. A Detailed operation of temporal confluent motion information candidate derivation unit 161 will be described later.

[00142] If the temporal candidate use permission flag is 0 (N in S111), S112 will be skipped.

[00143] Then, the confluent motion information candidate list construction unit 162 builds a confluent motion information candidate list from the spatial confluent motion information candidates from zero to the maximum amount of motion information candidates. spatial confluent motion and from temporal confluent motion information candidates from zero to the maximum amount of temporal confluent motion information candidates (S113). In this case, it is assumed that the spatial confluent motion information candidates and the temporal confluent motion information candidates are sequentially added to the list of confluent motion information candidates in the order of the derivation.

[00144] The redundant confluent movement information candidate deletion unit 163 then examines the confluent movement information candidates added to the confluent movement information candidate list supplied by the movement information candidate list construction unit 162, leave, if there is a plurality of confluent motion information candidates having the same motion information, one of the plurality of confluent motion information candidates while erasing the rest of the confluent motion information candidates (S114) and supplying the confluent movement information candidate list to the first confluent movement information candidate supply unit 164. The confluent movement information candidates added to the confluent movement information candidate list are all different confluent movement information candidates n this moment.

[00145] In this case, the redundant confluent movement information candidate deletion unit 163 completely deletes the confluent movement information candidates that have the same movement information from the confluent movement information candidate list. However, as long as the addition of a first supplemental confluent movement information candidate or a second supplementary confluent movement information candidate can be facilitated, the deletion is not limited to that.

[00146] The first confluent motion information candidate supply unit 164 then derives zero to the first two supplemental confluent motion information candidates from the confluent motion information candidates added to the confluent motion information candidate list supplied by the redundant confluent movement information candidate deletion unit 163 so as to add the first supplementary confluent movement information candidates to the list of confluent movement information candidates in such a way that the maximum amount of confluence candidates is not exceeded (S115) and supplies the confluent movement information candidate list to the second confluent movement information candidate supply unit 165. If the quantity of the confluent movement information candidates added to the confluent movement information candidate list a supplied by the redundant confluent movement information candidate deletion unit 163 is the maximum amount of confluence candidates, the confluent movement information candidate list supplied by the redundant confluent movement information candidate deletion unit 163 will be directly supplied to the second confluent movement information candidate supply unit 165. The detailed operation of the first confluent movement information candidate supply unit 164 will be described later.

[00147] The second confluent motion information candidate supply unit 165 then derives the second supplemental confluent motion information candidates in such a way that the amount of confluent motion information candidates added to the list of confluent motion information candidates confluent movement supplied by the first confluent movement information candidate supply unit 164 does not exceed the maximum quantity of confluence candidates so as to add the second supplementary confluent movement information candidates to the confluent movement information candidate list (S116) and supplies the confluent movement information candidate list to terminal 19. If the quantity of the confluent movement information candidates added to the confluent movement information candidate list supplied by the first movement information candidate supply unit 164 is the maximum amount of confluence candidates, the list of confluent movement information candidates supplied by the first confluent movement information candidate supply unit 164 will be supplied to the terminal 19. The detailed operation of the second candidate supply unit of confluent motion information 165 will be described later.

[00148] As described above, the confluent motion information candidate list building unit 140 builds a confluent motion information candidate list.

BLOCK GROUP

[00149] Hereinafter, an explanation is given with respect to a group of candidate blocks of a prediction block. Figures 15A and 15B are diagrams explaining a special candidate block group of a prediction block. Figure 15A shows a special candidate block group of a prediction block size that occurs when the size of a coding block is 16 pixels x 16 pixels and a prediction block size type is 2Nx2N.

[00150] The special candidate block group includes five blocks: an A block situated to the left of a lower left pixel of the prediction block; a block B located above an upper right pixel of the prediction block; a block C located diagonally above to the right of an upper right pixel of the prediction block; an E block located diagonally below left of a lower left pixel of the prediction block; and a D block located diagonally above left of an upper left pixel of the prediction block.

[00151] The group of temporal candidate blocks includes two blocks: an H block and an I block, which are representative blocks in a predetermined area of a ColPic. When the top left pixel position of the target prediction block is set to be (x,y) and the width and height of the target prediction block are set to be PUW and PUH respectively, a block in ColPic that includes a position pixel of ((((x + PUW) >> 4) << 4), (((S + PUH) >> 4) << 4)) as the top left pixel position of the block is defined to be a temporal candidate block H, where ">>" represents a bit shift in a right direction and "<<" represents a bit shift in a left direction.

[00152] Similarly, a block in ColPic that includes a pixel position of (x + (PUW >> 1), y + (PUH >> 1)) as the upper left pixel position of the block is set to be a temporal candidate block I.

[00153] It is observed that a positional relationship that is the same as that of a predictive block of a coding block that has a prediction block size type of 2Nx2N is applied to a predictive block in a coding block which does not have a prediction block size type of 2Nx2N. Figure 15B shows a special candidate block group that occurs when a coding block size is 8 pixels x 8 pixels and a prediction block size type is 2NxN.

[00154] In the Figure, block A is a block located to the left of a lower left pixel of the prediction block. However, as long as block A is in contact with the left side of the prediction block, the position of block A is not limited to that. Also, block B is a block located one pixel above the upper right of the prediction block. However, as long as block B is in contact with the upper side of the prediction block, the position of block B is not limited to that. The group of temporal candidate blocks is defined to include two blocks, the H block and the I block. However, since the group of temporal candidate blocks includes a block in a prediction block at the same position and a block around the block of prediction at the same position, the group of temporal candidate blocks is not limited to this.

DETAILED OPERATION OF SPATIAL CONFLUENT MOTION INFORMATION CANDIDATE DERIVATION UNIT 160

[00155] An explanation is now given on the detailed operation of the spatial confluent motion information candidate derivation unit 160. Figure 16 is a flowchart for explaining the operation of the spatial confluent motion information candidate derivation unit 160. The spatial confluent movement information candidate derivation unit 160 repeats the following processes in the order of an A block, a B block, a C block, an E block and a D block, which are candidate blocks included in a group of blocks of a special candidate block group (S120 to S124).

[00156] First, the spatial confluent motion information candidate derivation unit 160 checks whether a candidate block is valid (S121). A candidate block that is valid means that at least one of the respective L0 prediction and L1 prediction reference indices of the candidate block is greater than or equal to 0. If the candidate block is valid (S at S121), the derivation unit Spatial confluent motion information candidate block 160 will derive the candidate block motion information (S122). If the candidate block is not valid (N at S121), the spatial confluent motion information candidate derivation unit 160 will check a subsequent candidate block (S124) and terminate the process upon completion of checking all candidate blocks. Subsequently to step S122, the spatial confluent motion information candidate derivation unit 160 checks whether the amount of derived spatial confluent motion information candidates is the maximum amount of spatial confluent motion information candidates (S123). If the amount of derived spatial confluent motion information candidates is not the maximum amount of spatial confluent motion information candidates (N at S123), the spatial confluent motion information candidate derivation unit 160 checks a subsequent candidate block ( S124). If the amount of derived spatial confluent motion information candidates is the maximum amount of spatial confluent motion information candidates (S at S123), the spatial confluent motion information candidate derivation unit 160 will terminate the process. The maximum amount of spatial confluent motion information candidates is set to be 4.

DETAILED OPERATION OF CONFLUENT TEMPORAL MOVEMENT INFORMATION CANDIDATE DERIVATION UNIT 161

[00157] An explanation is now given on the detailed operation of the temporal confluent motion information candidate derivation unit 161. Fig. 17 is a flowchart for explaining the operation of the temporal confluent motion information candidate derivation unit 161. The temporal confluent movement information candidate derivation unit 161 repeats the following processes for each L0 forecast and L1 forecast LX forecast direction (S140 to S147). X is 0 or 1. The temporal confluent movement information candidate derivation unit 161 repeats the following processes in the order of an H block and an I block, which are candidate blocks included in a group of temporal candidate blocks of a group of candidate blocks (S141 to S146).

[00158] Temporal confluent movement information candidate derivation unit 161 checks whether the LN prediction of a candidate block is valid (S142). N is 0 or 1. It is assumed that N is the same as X. The LN prediction of a candidate block being valid means that a reference index of the LN prediction of the candidate block is greater than or equal to 0. If the LN prediction of candidate block is valid (S at S142), the temporal confluent motion information candidate derivation unit 161 will set a prediction motion vector LN of the candidate block to be a prediction reference motion vector LX(S143). If the prediction LN of the candidate block is not valid (N at S142), the temporal confluent movement information candidate derivation unit 161 will skip steps S143 to S145 and check a subsequent candidate block (S146). Upon completion of verification of all candidate blocks, temporal confluent motion information candidate derivation unit 161 processes subsequent prediction direction (S147).

[00159] Subsequent to the step S143, the temporal confluent motion information candidate derivation unit 161 determines a reference representation of prediction LX of a temporal confluent motion information candidate (S144). In this case, the reference representation of the prediction LX of the temporal confluent motion information candidate is defined to be a reference representation of a reference index 0. Then, scaling the reference motion vector to match a distance between the target representation and the reference representation of the prediction LX of the temporal confluent motion information candidate, the temporal confluent motion information candidate derivation unit 161 calculates a motion vector of the prediction LX of the confluent motion information candidate (S145) and processes the subsequent forecast direction (S147). A specific method of calculating the motion vector of the candidate LX prediction of temporal confluent motion information will be described later.

[00160] Subsequent to the step S147 in which the processes are finalized for the forecast L0 and the forecast L1, the temporal confluent movement information candidate derivation unit 161 checks whether at least one of the forecast L0 and the forecast L1 of the candidate of temporal confluent motion information is valid (S148). If at least one of the forecast L0 and forecast L1 of the temporal confluent movement information candidate is valid (S at S148), the temporal confluent movement information candidate derivation unit 161 determines the interforecast type of the temporal confluent movement information candidate. temporal confluent motion and derive the candidate from temporal confluent motion information (S149).

[00161] For the determination of the interforecast type, the interforecast type of the temporal confluent movement information candidate is defined to be Pred_L0 if only the L0 forecast is valid, the interforecast type of the temporal confluent movement information candidate is defined to be Pred_L1 if only the L1 forecast is valid and the temporal confluent movement information candidate interforecast type is set to be Pred_BI if both the L0 forecast and the L1 forecast are valid.

[00162] Subsequently, an explanation of the motion vector calculation method of the LX prediction of the temporal confluent motion information candidate is given. If an inter-representation distance between a ColPic that has a temporal candidate block and a ColRefLXPic that is a representation referred to by the temporal candidate block in the LX prediction motion compensation prediction, an inter-representation distance between a reference picture RefLXPic of the temporal confluent motion information candidate LX prediction and a target representation CurPic and the LX prediction reference motion vector are denoted as td, tb and mvLX, respectively, a motion vector mvLXCol of the motion information candidate LX prediction temporal confluent motion will be calculated through Expression 1. It can be understood based on Expression 1 that subtraction, division and multiplication to calculate tb and td are necessary to calculate the motion vector of the candidate LX prediction of temporal confluent motion information . mvLXCol = tb / td * mvLX Expression 1

[00163] In the case of using integer arithmetic for the simplification of floating point arithmetic, for example, the expression can be used after being expanded as in Expression 2 up to Expression 4. Abs(v) is a function to calculate the absolute value of a v value. Clip3(uv,lv,v) is a function that limits the value v to be from a lower bound lv to an upper bound uv. Sign(v) is a function that returns 1 if the value v is greater than or equal to 0 and returns -1 if the value v is less than 0, where ">>" represents a bit shift in a direction to right and "<<" represents a bit shift in a left direction. tx = (16384 + Abs(td / 2)) / td Expression 2 DistScaleFactor = Clip3(-4096, 4095, (tb * tx + 32) >> 6); Expression 3 mvLXCol = Sign( DistScaleFactor * mvLX ) * ( (Abs( DistScaleFactor * mvLX ) + 127 ) >> 8)) Expression 4

[00164] In this case, the maximum amount of temporal confluent movement information candidates is set to be 1. Therefore, although a process that corresponds to step S123 shown in Figure 17, which is a flowchart that explains the operation of the bypass unit of spatial confluent motion information candidate 160, is omitted in Fig. 17 , the process corresponding to step S123 may be added after step S149 if the maximum amount of temporal confluent motion information candidates is greater than or equal to 2.

[00165] In this case, the interforecast type of the temporal confluent movement information candidate varies. Alternatively, for example, a dependency relationship between X and N can be removed so that the check is performed in the order of N being 0 and 1 for X so as to always generate Pred_BI.

DETAILED OPERATION OF THE FIRST CONFLUENT MOVEMENT INFORMATION SUPPLY UNIT 164

[00166] An explanation is now given on the detailed operation of the first confluent movement information candidate supply unit 164. Figure 18 is a flowchart for explaining the operation of the first confluent movement information candidate supply unit 164. First, the first confluent motion information candidate supply unit 164 calculates MaxNumGenCand, which is the maximum amount for deriving the first supplemental confluent motion information candidates, through Expression 5 from the amount of confluent motion information candidates (NumCandList) and the maximum amount of confluence candidates (MaxNumMergeCand) added to the confluent move information candidate list (S170). MaxNumGenCand = MaxNumMergeCand - NumCandList; (NumCandList > 1) MaxNumGenCand = 0; (NumCandList <= 1) Expression 5

[00167] Then, the first confluent motion information candidate supply unit 164 checks whether MaxNumGenCand is greater than 0 (S171). If MaxNumGenCand is not greater than 0 (N at S171), the first confluent motion information candidate supply unit 164 will terminate the process. If MaxNumGenCand is greater than 0 (S at S171), the first confluent motion information candidate supply unit 164 performs the following processes. First, the first confluent movement information candidate supply unit 164 determines loopTimes, which is the amount of combination checks. Set loopTimes to be NumCandListxNumCandList. If loopTimes exceeds 8, loopTimes is limited to 8 (S172). loopTimes is an integer from 0 to 7. The following processes are repeated for loopTimes only (S172 through S180). The first confluent motion information candidate supply unit 164 determines a combination of a confluent motion information candidate M and a confluent motion information candidate N (S173). The relationships between the amount of combination checks, the confluent movement information candidate M, and the confluent movement information candidate N. Figure 19 is a diagram to explain relationships between the quantity of combination checks, the movement information candidate confluent motion information M and the confluent motion information candidate N. As in Figure 19, M and N are different values and are defined in ascending order of the total value of M and N. The first confluent motion information candidate supply unit 164 checks whether the L0 prediction of the confluent movement information candidate M is valid and whether the L1 prediction of the confluent movement information candidate N is valid (S174). If the confluent movement information candidate forecast L0 is valid and the confluent movement information candidate forecast L1 is valid (S at S174), the first confluent movement information candidate supply unit 164 checks whether a reference representation and a motion vector of L0 prediction of confluent motion information candidate M are different from a reference representation and a motion vector of L1 prediction of confluent motion information candidate N (S175). If the forecast L0 of the candidate of confluent movement information M is valid and the prediction L1 of the candidate of confluent movement information N is not valid (N in S174), the first supplying unit of confluent movement information candidate 164 processes a subsequent combination. If the L0 forecast reference representation of the confluent movement information candidate M is different from the L1 forecast reference representation of the confluent movement information candidate N is valid (S in S175), the first information candidate supply unit confluent motion information candidate 164 derives a bi-confluent motion information candidate that has an interforecast type of Pred_BI by combining the motion vector and the prediction reference representation L0 of the confluent motion information candidate M and the motion vector and the prediction reference representation L1 of the confluent movement information candidate N (S176). In that case, the first confluent movement information candidate supply unit 164 derives, as a supplementary first confluent movement information candidate, the bi-confluent movement information obtained by combining the L0 prediction movement information of a candidate of confluent motion information and motion information from the L1 prediction of a different confluent motion information candidate. If the L0 prediction reference representation of the confluent movement information candidate M is the same as the L1 forecast reference representation of the confluent movement information candidate N (N in S175), the first information candidate supply unit of confluent motion 164 processes a subsequent combination. Subsequent to step S176, the first confluent movement information candidate supply unit 164 adds the bi-confluent movement information candidate to the confluent movement information candidate list (S178). Subsequently to step S178, the first confluent motion information candidate supply unit 164 checks whether the chunk quantity of the derived bi-confluent motion information is MaxNumGenCand (S179). If the quantity of the derived bi-confluent motion information pieces is MaxNumGenCand (S at S179), the first confluent motion information candidate supply unit 164 will terminate the process. If the quantity of the derived bi-confluent motion information pieces is not MaxNumGenCand (N at S179), the first confluent motion information candidate supply unit 164 processes a subsequent combination (S180). The first confluent movement information candidate supply unit 164 ends the process when the processing of all combinations is finished.

[00168] In this case, the first supplemental confluent motion information candidate is defined to be a bi-confluent motion information candidate, where the motion compensation prediction direction is bidirectional by combining the motion vector and the L0 prediction reference representation of a confluent motion information candidate added to the confluent motion information candidate list, and the motion vector and the L1 prediction reference representation of another confluent motion information candidate. However, the first candidate of supplemental confluent movement information is not limited to this. In this case, if NumCandList is less than or equal to 1, the motion vector and L0 prediction reference representation of a confluent motion information candidate added to the confluent motion information candidate list cannot be combined with the motion vector and the L1 prediction reference representation of another confluent motion information candidate. Thus, MaxNumGenCand is set to be 0. For example, the first supplemental confluent motion information candidate could be a confluent motion information candidate, where the motion compensation prediction direction is bidirectional, obtained by adding a offset value of +1 or similar to the L0 prediction motion vector and the L1 prediction motion vector of a confluent motion information candidate added to the confluent motion information candidate list or a targeted confluent motion information candidate to the double prediction obtained by adding an offset value of +1 or similar to the L0 prediction motion vector or the L1 prediction motion vector of a confluent motion information candidate added to the confluent motion information candidate list.

[00169] In this case, if there is a slight difference between the movement information of a confluent movement information candidate added to the confluent movement information candidate list and the movement of a target movement information candidate, the coding efficiency can be augmented by deriving a new confluent motion information candidate, where the motion compensation prediction direction is bidirectional, by combining the motion vector and L0 prediction reference representation of a motion information candidate confluent added to the list of confluent motion information candidates and the motion vector and the L1 prediction reference representation of another confluent motion information candidate. DETAILED OPERATION OF THE SECOND CONFLUENT MOVEMENT INFORMATION SUPPLY UNIT 165

[00170] An explanation of the detailed operation of the second confluent movement information candidate supply unit 165 is now given. Figure 20 is a flowchart for explaining the operation of the second confluent movement information candidate supply unit 165. In that case, an explanation is given on the condition that the amount of reference representations available in the reference representation list L0 is greater than or equal to the amount of reference representations available in the reference representation list L1.

[00171] First, the second confluent motion information candidate supply unit 165 calculates MaxNumGenCand, which is the maximum amount for deriving the second supplementary confluent motion information candidates, through Expression 6 from the amount of information candidates of confluent movement information (NumCandList), the maximum number of confluence candidates (MaxNumMergeCand), and the number of available reference representations in the L0 reference representation list (num_ref_idx_l0_active_minus1+1) added to the candidate list of confluent movement information supplied by the first confluent movement information candidate supply unit 164 (S180). In the expression, Min(x,y) is a function that returns the respective minimum values of x and y. MaxNumGenCand = Min(MaxNumMergeCand - NumCandList, num_ref_idx_l0_active_minus1+1) Expression 6

[00172] Then, the second confluent movement information candidate supply unit 165 repeats the following processes for the number of times from MaxNumGenCand to i (S181 to S186), where i is an integer from 0 to (MaxNumGenCand - 1). In other words, if the number of confluent motion information candidates (NumCandList) added to the confluent motion information candidate list supplied by the first confluent motion information candidate supply unit 164 is less than the maximum number of candidates (MaxNumMergeCand), at least a second supplementary confluent motion information candidate will be derived and added to the list of confluent motion information candidates. On the other hand, if NumCandList is greater than or equal to MaxNumMergeCand, a second supplemental confluent motion information candidate is not derived and a second supplemental confluent motion information candidate is not added to the confluent motion information candidate list. As described, if NumCandList is greater than or equal to MaxNumMergeCand, a memory for a candidate list of confluent motion information where NumCandList is greater than MaxNumMergeCand becomes unnecessary by not adding a second supplemental candidate of confluent motion information to the list of confluent movement information candidates. The second confluent motion information candidate supply unit 165 derives a second supplementary confluent motion information candidate having a motion vector (mvL0) of (0,0) for prediction L0, a reference index (ref_idx_l0) of i for forecast L0 and an interforecast type (predDir) of Pred_L0 (S182).

[00173] Subsequently, the second confluent movement information candidate supply unit 165 checks whether the slice type is slice B (S183). If the slice type is slice B (S at S183), the second confluent motion information candidate supply unit 165 will derive a second supplementary confluent motion information candidate having a motion vector (mvL1) of (0 ,0) for forecast L1, a reference index (ref_idx_l1) of 0 for forecast L1, and an interforecast type of Pred_BI (S184) and will add the second supplemental confluent movement information candidate to the movement information candidate list confluent (S185).

[00174] If the slice type is not slice B (N at S183), the second confluent movement information candidate supply unit 165 will add the derived second supplementary confluent movement information candidate at S182 to the information candidate list of confluent movement (S185).

[00175] Subsequently, the second confluent movement information candidate supply unit 165 processes a subsequent i (S186). The second confluent movement information candidate supply unit 165 ends the process when the processing of all the i's is finished.

[00176] In this case, the second supplemental confluent motion information candidate is defined to be a confluent motion information candidate that has a motion vector of (0,0) and a reference index of i for prediction L0 and an interforecast type of Pred_L0 in the case of a slice P and the second supplementary confluent motion information candidate is defined to be a confluent motion information candidate that has a motion vector of (0,0) and a reference index of i for forecast L0, a motion vector of (0,0) and a reference index of 0 for forecast L1, and an interforecast type of Pred_BI in the case of a slice B.

[00177] This is due to the fact that in a commonly used moving representation where a part of a representation is stationary, a motion vector of (0,0) is valid and the correlation with a target representation becomes more high as a reference representation becomes closer to the target representation since a moving representation is continuous in a temporal direction.

[00178] A detailed explanation will be given in the following regarding the second candidate of supplementary confluent movement information in the case of a slice B.

[00179] First, an explanation is given regarding encoding without rearranging between the order of an input representation and the encoding order and regarding encoding with rearranging between the order of an input representation and the encoding order.

[00180] Figures 21A and 21B are diagrams explaining a relationship between a target representation and a reference representation. Figure 21A shows a case where there is no rearrangement between the order of an input representation and the encoding order and Figure 21B shows a case where there is rearrangement between the order of an input representation and the encoding order. In Figure 21A, representations having a POC of n+4 to n+7 represent the reference representations, and a representation having a POC of n+8 represents a CurPic target representation. In Figure 21B, representations that have a POC of n+2, n+4, n+6, or n+8 represent the reference representations, and a representation that has a POC of n+5 represents a target CurPic representation.

[00181] Next, an explanation is given regarding a list of reference representation in encoding in which there is no rearrangement between the order of an input representation and the order of encoding and a list of reference representation in encoding in which there is rearrangement between the order of an input representation and the order of encoding. A reference representation list is built by the reference representation list construction unit 113.

[00182] Figures 22A and 22B are diagrams that explain an example of a reference representation list. Figs. 22A and 22B show an example where the quantity of available reference representations in an L0 reference representation list and the quantity of available reference representations in an L1 reference representation list are both 4. Fig. 22A shows the condition of a reference representation list when there is no rearrangement between the order of an input representation and the encoding order. Assuming that at least one reference representation that has a POC of n+4, one reference representation that has a POC of n+5, one reference representation that has a POC of n+6, and one reference representation that has a POC of n+8 are stored in frame memory 110. Figure 22B shows the condition of a reference representation list when there is rearrangement between the order of an input representation and the order of encoding. Assuming that at least one reference representation that has a POC of n+2, one reference representation that has a POC of n+4, one reference representation that has a POC of n+6, and one reference representation that has a POC of n+8 are stored in the frame memory 110. In the case where there is no rearrangement between the order of an input representation and the encoding order, a reference representation list L0 and a reference representation list L1 are identical. In the case where there is rearrangement between the order of an input representation and the coding order, an L0 reference representation list and an L1 reference representation list are not identical.

[00183] Next, an explanation is given regarding a second candidate of supplementary confluent motion information in encoding in which there is no rearrangement between the order of an input representation and the encoding order, and a second candidate of motion information supplementary confluent in encoding where there is no rearrangement between the order of an input representation and the order of encoding.

[00184] Figures 23A and 23B are diagrams explaining an example of a relationship between a reference index of a second candidate of supplemental confluent movement information and a POC according to the first embodiment. Figs. 23A and 23B show an example of a second supplemental confluent motion information candidate when the quantity of reference representations available in a reference representation list L0 and the quantity of reference representations available in a reference representation list L1 are both 4. Figure 23A shows a reference index of a second candidate of supplemental confluent movement information when there is no rearrangement between the order of an input representation and the order of encoding. Figure 23B shows a reference index of a second candidate of supplemental confluent movement information when rearranging between the order of an input representation and the coding order.

[00185] An explanation is given regarding the case of Figure 23A. The reference index of a 0th (i = 0) second supplemental confluent movement information candidate for the L0 prediction is 0 and a reference representation that has a POC of n+7 is therefore used for the prediction motion compensation for the L0 forecast. The reference index for the L1 forecast is 0 and a reference representation that has a POC of n+7 is therefore used for the motion compensation forecast for the L1 forecast. The prediction motion vector L0 and the prediction motion vector L1 are identical being at (0,0) and the respective reference representations are also identical. Therefore, the respective forecast signals of the L0 forecast and the L1 forecast are also identical. Therefore, although the 0-th second supplemental confluent motion information candidate has motion information for dual prediction, the prediction signal obtained is identical to a prediction signal for uni-prediction obtained by performing motion compensation prediction. in a reference representation that has a POC of n+7 using a motion vector of (0,0).

[00186] In a first second candidate of supplementary confluent motion information, where a reference representation has a POC of n+6 is used for the motion compensation prediction for the L0 prediction and a reference representation that has a POC of n+7 is used for motion compensation prediction for L1 prediction. This represents the dual-forecast motion compensation forecast where the nearest (closest) reference representation and the second closest reference representation are used.

[00187] Similarly, in a second second candidate of supplemental confluent motion information, dual prediction motion compensation prediction is performed in which the closest reference representation and the third closest reference representation are used. In a third, second candidate of supplemental confluent motion information, dual prediction motion compensation prediction is performed in which the nearest reference representation and the fourth closest reference representation are used.

[00188] An explanation is given regarding the case of Figure 23B. The reference index of a 0th (i = 0) second supplemental confluent movement information candidate for the L0 prediction is 0 and a reference representation that has a POC of n+4 is therefore used for the prediction motion compensation for the L0 forecast. The reference index for the L1 forecast is 0 and a reference representation that has a POC of n+6 is therefore used for the motion compensation forecast for the L1 forecast. The reference representation that has a POC of n+4 and the reference representation that has a POC of n+6 are a reference representation immediately before the target representation and a reference representation immediately after the target representation in the display order, respectively. In this way, dual prediction motion compensation prediction is performed which is considered to have the highest relative selectivity among the second supplemental confluent motion information candidates.

[00189] In a first second candidate of supplementary confluent motion information, where a reference representation has a POC of n+2 is used for the motion compensation prediction for the L0 prediction and a reference representation that has a POC of n+6 is used for motion compensation prediction for L1 prediction. This represents dual prediction motion compensation prediction in which a second reference representation before the target representation and a reference representation immediately after the target representation in the display order are used.

[00190] Although the second supplemental confluent motion information candidate has the motion information for the dual prediction, the prediction signal obtained is identical to a prediction signal for the uni-prediction obtained by performing the motion compensation prediction in a reference representation that has a POC of n+6 using a motion vector of (0,0).

[00191] In a third second candidate of supplementary confluent motion information, where a reference representation has a POC of n+8 is used for motion compensation prediction for the L0 prediction and a reference representation that has a POC of n+6 is used for motion compensation prediction for L1 prediction. This represents dual prediction motion compensation prediction in which a second reference representation after the target representation and a reference representation immediately after the target representation in the display order are used.

[00192] As described above, all second supplemental confluent motion information candidates can be driven to the dual prediction where the nearest reference representation is used by defining the motion vector of the L0 prediction of a second candidate of motion information supplementary confluent motion of a slice B to be (0,0), which sets a motion vector of prediction L1 to be (0,0) which increments the reference index of prediction L0 by one at a time, and which defines the index L1 prediction reference representation to be 0 where the closest reference representation in L1 prediction reference representation list is added.

[00193] Also, if there is no reference representation for prediction L1, allowing a second supplemental confluent motion information candidate to be directed to the uni-forecast from prediction L0, selections of a second information candidate additional confluent motion data can be increased and the prediction efficiency of a motion information prediction process can thereby be improved.

[00194] Also, since every second candidate of supplemental confluent motion information represents the different prediction signals, the coding efficiency can be improved by expanding the range of choices for a confluence mode.

[00195] Also, second candidates of supplementary confluent movement information as well as the quantity of reference representations available in the reference representation list L0 can be derived without depending on the quantity of reference representations available in the reference representation list L1.

[00196] For example, in a moving representation where one part of a representation is stationary with random noise mixing, using a prediction signal of the dual prediction where the reference representation closer and another representation of reference are used, a prediction signal may be able to be obtained whose accuracy is higher than that of a uni-prediction prediction signal in which only the closest reference representation is used relative to the sharpness of each reference representation. reference through a filtering effect and a quantization width.

[00197] Defining a confluent movement information candidate that does not depend on a confluent movement information candidate added in the confluent movement information candidate list as the second supplementary confluent movement information candidate allows the use of a mode of confluence information when the amount of confluent movement information candidates added in the confluent movement information candidate list is zero and the coding efficiency can be improved. Also, when the movement information of a confluent movement information candidate added to the confluent movement information candidate list is different from the movement of the target candidate movement information, the coding efficiency can be improved by increasing the choices derived from it. if a new candidate of confluent movement information.

[00198] Also, by applying all second supplemental confluent motion information candidates to be directed to the dual prediction, the probability of a first supplemental confluent motion information candidate in an adjacent block to be derived can be increased and the efficiency encoding can thus be improved. A detailed explanation regarding their effects is given below.

[00199] First, an explanation is given regarding a problem that occurs when a second supplemental confluent motion information candidate targeting the uni-forecast of forecast L0 is added to the list of confluent motion information candidates. If a first supplementary confluent motion information candidate is derived by combining a second supplementary confluent motion information candidate of uni-forecast of the L0 forecast and a predetermined confluent motion information candidate in the list of confluent motion information candidates, a new second supplemental confluent motion information candidate will not be derived if the predetermined confluent motion information candidate is targeted to the L0 forecast uni-forecast. Even when the predetermined confluent motion information candidate is directed to the dual forecast, once a motion vector and a reference representation of the L1 prediction that must be matched with those of the L0 prediction of the predetermined confluent motion information candidate do not exist in the second supplemental confluent motion information candidate, a first supplementary confluent motion information candidate using the prediction L0 of the predetermined confluent motion information candidate is not derived.

[00200] In this case, by allowing all second supplemental confluent motion information candidates to be directed to the dual prediction, the derivation of more first supplementary confluent motion information candidates may be possible when combined with a motion information candidate predetermined confluent in the candidate list of confluent movement information, and the coding efficiency can thereby be improved.

[00201] In this case, the special candidate block group is defined to have five blocks, the temporal candidate block group is defined to have two blocks, the upper limit of the maximum amount of confluence candidates is defined to be 5, the maximum amount of spatial confluent motion information candidates is set to be 4 and maximum amount of temporal confluent motion information candidates is set to be 1. However, the respective amounts are not limited to this.

[00202] In this case, the respective motion vectors of forecast L0 and forecast L1 are defined to be (0,0). However, motion vectors are not limited to this, as long as motion vectors do not depend on the motion information of a confluent motion information candidate added to the confluent motion information candidate list, and have a statistically high frequency of use. . For example, a motion vector of the second candidate of supplemental confluent motion information can be defined to be a motion vector that has a high frequency of occurrence of a coded representation or a portion of a coded representation so as to be coded in a bit stream and then transmitted. The reference index of forecast L1 is set to be 0. However, if there is the nearest reference representation which is beyond 0, the reference index can be fixed to a reference index beyond 0.

[00203] Also, since the amount of reference representations available in the reference representation list L0 often becomes greater than or equal to the amount of reference representations available in the reference representation list L1 generally, the amount of representations reference index available in the reference representation list L0 is used for the calculation of the maximum number of second candidates of supplementary confluent movement information to be derived and the reference index of prediction L1 is fixed while changing the reference index of prediction L0 in that case. However, when the amount of reference representations available in the reference representation list L1 is greater than the amount of reference representations available in the reference representation list L0, the amount of reference representations available in the reference representation list L1 may be used for calculating the maximum amount of candidate seconds of supplemental confluent movement information to be derived, and the prediction reference index L0 can be fixed while changing the prediction reference index L1.

VECTOR PREDICTION MODE DETERMINATION UNIT CONFIGURATION 120

[00204] Subsequently, an explanation is given regarding the detailed configuration of the vector prediction mode determination unit 120. Fig. 24 is a diagram showing the configuration of the vector prediction mode determination unit 120. The unit a vector prediction mode determining unit 120 includes a vector prediction candidate list construction unit 130 and a vector prediction prediction determining unit 131. A terminal 17 is connected to the prediction coding mode determining unit 122 .

[00205] The vector prediction candidate list construction unit 130 is likewise provided in the motion vector reproduction unit 211 inside the motion representation decoding device 200 which decodes a bit stream generated through of a motion representation encoding device 100 according to the first embodiment, and a list of identical vector prediction candidates is constructed each in the motion representation encoding device 100 and the motion representation decoding device 200.

OPERATION OF VECTOR PREDICTION MODE DETERMINATION UNIT 120

[00206] An explanation is given below regarding the operation of the vector prediction mode determination unit 120. Figure 25 is a flowchart showing the operation of the vector prediction mode determination unit 120.

[00207] First, the following processes are performed for the L0 forecast. In the following, x represents 0. The vector prediction candidate list construction unit 130 derives a prediction reference index LX supplied by the terminal 13. The vector prediction candidate list construction unit 130 builds a list of prediction vector candidates of the LX prediction which includes the prediction vector candidates of the maximum amount of prediction vector candidates from the candidate block group supplied by the terminal 12 and the reference index of the LX prediction (S200). The vector forecast candidate list construction unit 130 supplies the forecast vector list candidate list LX to the vector forecast determination unit 131.

[00208] The vector forecast determination unit 131 selects a vector forecast candidate from the vector forecast candidate list of the LX forecast supplied by the vector forecast candidate list construction unit 130 (S201) and determines a vector forecast index from the LX forecast (S202).

[00209] The vector forecast determination unit 131 calculates a vector difference of the forecast LX by subtracting the vector forecast of the forecast LX from the motion vector of the forecast LX supplied by the terminal 13 (S203) and outputs the vector difference of the LX forecast and the vector forecast index of the LX forecast.

[00210] The vector forecast determination unit 131 calculates a forecast error amount from the representation signal supplied by the terminal 15 and from a forecast forecast signal LX obtained by performing the motion compensation forecast in the reference representation supplied by terminal 14 based on the forecast motion vector LX and the forecast reference index LX supplied by terminal 13 and calculates a rate distortion evaluation value of Pred_LX from the amount of forecast error and from the encoding amount of the LX forecast vector difference, the LX forecast reference index, and the LX forecast vector index.

[00211] Then, a process that is the same as that for prediction L0 is performed for prediction L1 while setting X to be 1.

[00212] Subsequently, the vector prediction determining unit 131 calculates a prediction error amount from the representation signal supplied by the terminal 15 and from a prediction prediction signal BI obtained by averaging the signal prediction value of prediction L0 and prediction signal of prediction L1 and calculates a rate distortion evaluation value of Pred_BI from the amount of prediction error and from the encoding amount of the respective vector differences of prediction L0 and prediction L1 forecast, the respective reference key figures of L0 forecast and L1 forecast, and the vector forecast key figures of L0 forecast and L1 forecast.

[00213] The vector prediction determination unit 131 compares the rate distortion assessment value of Pred_L0, the rate distortion assessment value of Pred_L1 and the rate distortion assessment value of Pred_BI and selects a mode of prediction encoding with the lowest rate distortion assessment value. The vector prediction determination unit 131 then supplies the motion information, vector differences, vector prediction indices and rate distortion assessment value based on the prediction coding mode to the prediction coding unit. prediction coding mode determination 122. If prediction coding mode is Pred_L0, motion vector of prediction L1 is set to be (0,0) and reference index of prediction L1 is set to be "-1 ". If the forecast coding mode is Pred_L1, the L0 forecast motion vector is set to be (0,0) and the L0 forecast reference index is set to be "-1". VECTOR PREDICTION CANDIDATE LIST BUILDING UNIT CONFIGURATION 130

[00214] An explanation is given, then, about the detailed configuration of the vector prediction candidate list construction unit 130. Figure 26 is a diagram for explaining the configuration of the vector prediction candidate list construction unit 130. A terminal 18 is connected to the vector prediction determination unit 131. The vector prediction candidate list construction unit 130 includes a space vector predictor candidate derivation unit 150, a candidate derivation unit 130. scale-scaled scaling prediction vector prediction unit 151, a temporal vector prediction candidate derivation unit 152, a vector prediction list building unit 153, a redundant vector prediction candidate deletion unit 154 and a vector prediction candidate supply unit 155.

VECTOR PREDICTION CANDIDATE LIST BUILDING UNIT OPERATION 130

[00215] An explanation is given below regarding the function and operation of each component. Figure 27 is a flowchart for explaining the operation of the vector prediction candidate list building unit 130.

[00216] The vector prediction candidate list construction unit 130 separates the candidate blocks included in the special candidate block group supplied by the terminal 12 into two groups: An E block and an A block that form a first group; and a C block, a B block and a D block which form a second group and repeat the following processes in order of the first group and the second group (S210 to S212).

[00217] The candidate block group supplied by terminal 12 is a candidate block group which is the same as the one in the confluence mode case.

[00218] The spatial vector predictor candidate derivation unit 150 derives zero or one spatial vector prediction candidate of the LX prediction from the special candidate block group of an i-th group (i is 1 or 2) and the prediction reference index LX supplied by terminal 13 (S211) and supplies the prediction spatial vector prediction candidate LX to the vector prediction list building unit 153. The detailed operation of the predictor candidate derivation unit of space vector 150 will be described later. A relationship between the spatial scale prediction vector prediction candidate derivation unit 150 and the spatial scale scaling prediction vector prediction candidate derivation unit 151 will also be described later.

[00219] When the processes for the first group and the second group are finished, the vector prediction candidate list construction unit 130 checks whether the temporal candidate use permission flag (enable_temporal_mvp_flag) is 1 (S213).

[00220] If the temporal candidate use permission flag is 1 (S in S213), the temporal vector prediction candidate derivation unit 152 will derive zero or one temporal vector prediction candidate of the prediction LX from the group of temporal candidate blocks supplied by terminal 12 and the prediction reference index LX supplied by terminal 13 (S214) and supplies the temporal vector prediction candidate of prediction LX to vector prediction list building unit 153. The detailed operation time vector prediction candidate derivation unit 152 will be described later.

[00221] If the temporal candidate use permission flag is 0 (N in S213), step S214 will be skipped.

[00222] Subsequently, the vector prediction list building unit 153 builds a vector prediction candidate list of the LX prediction from zero to two special vector prediction candidates of the LX prediction and zero to one prediction candidate forecast vector list LX (S215) and supplies the list of forecast vector candidates LX to the redundant vector forecast candidate deletion unit 154. In this case, up to three vector forecast candidates are added to the forecast vector list. LX forecast vector prediction candidates from the spatial vector prediction candidates to the temporal vector prediction candidate, in order.

[00223] Subsequently, the redundant vector prediction candidate deletion unit 154 checks the vector prediction candidates added to the LX prediction vector prediction candidate list, leaves, when there are a plurality of vector prediction candidates that takes an identical vector, a vector prediction candidate, and deletes the rest of the vector prediction candidates. Then, the redundant vector prediction candidate deletion unit 154 deletes, when the quantity of vector prediction candidates added to the LX prediction vector prediction candidate list exceeds the maximum amount of vector prediction candidates, the candidates of the vector prediction candidates in the last part of the LX forecast vector prediction candidate list such that the quantity of the vector prediction candidates added to the LX forecast vector prediction candidate list becomes less than or equal to the quantity forecast vector candidate list (S216) and supplies the vector forecast candidate list of the forecast LX to the vector forecast candidate supply unit 155. Therefore, the vector forecast candidates added to the forecast candidate list The prediction vector data LX outputted through the redundant vector prediction candidate deletion unit 154 are all different prediction vector candidates. Also, when two special vector prediction candidates of LX prediction are left, a temporal vector prediction candidate of LX prediction is not added to the list of vector prediction candidates accordingly.

[00224] Subsequently, the vector prediction candidate supply unit 155 generates the vector prediction supplying the candidates, adds the vector prediction supplying candidates to the vector prediction candidate list of the prediction LX so that the number of vector prediction candidates added to the LX prediction vector prediction candidate list becomes the maximum amount of vector prediction candidates (S217) and supplies the vector prediction supplying candidates to terminal 18.

[00225] Assuming that the prediction vector that supplies the candidates has a motion vector (0,0) in this case. In this case, the prediction vector that supplies the candidates are defined to have a motion vector (0,0). However, the vector prediction that supplies the candidates can have a predetermined value, for example (1,1), it can be transmitted in a bitstream, or it can have a motion vector where a horizontal component or a vertical component of a spatial vector prediction candidate is defined to be either +1 or -1.

DETAILED OPERATION OF SPACE VECTOR PREDICTOR CANDIDATE DERIVATION UNIT 150

[00226] An explanation is now given on the detailed operation of the spatial vector predictor candidate derivation unit 150. Fig. 28 is a flow chart explaining the operation of the spatial vector predictor candidate derivation unit 150. an explanation regarding the derivation of a spatial vector prediction candidate from the LX prediction.

[00227] The space vector predictor candidate derivation unit 150 repeats the following processes for a candidate block in the i-th group (i is 1 or 2) (S220 to S226). The first group is scanned from E block to A block in order and the second group is scanned from C block to B block to D block in order.

[00228] The following processes are repeated for each candidate block in order of prediction L0 and prediction L1 (S221 to S225). Henceforth, an explanation is given while specifying the L0 forecast and the L1 forecast as the LN forecast. The space vector predictor candidate derivation unit 150 checks whether the candidate block prediction LN is valid (S222). The LN prediction of a candidate block being valid means that a reference index of the LN prediction of the candidate block is greater than or equal to 0.

[00229] If the prediction LN of the candidate block is valid (S at S222), the space vector predictor candidate derivation unit 150 checks whether a reference representation indicated by the prediction reference index LN of the candidate block is the same that a reference representation indicates by the forecast reference index LX (S223).

[00230] If the reference representations are the same (S at S223), the space vector predictor candidate derivation unit 150 will define a motion vector of the prediction LN of the candidate block to be a space vector prediction candidate of the LX forecast (S224) and terminate the process.

[00231] If the candidate block prediction LN is not valid (N at S222) or the reference representations are not the same (N at S223), the space vector predictor candidate derivation unit 150 will check the prediction L1 if LN prediction is L0 prediction and checks a subsequent candidate block if LN prediction is L1 prediction (S225).

[00232] Upon completion of verification of all candidate blocks (S226), the spatial vector predictor candidate derivation unit 150 checks whether the temporal candidate usage permission flag (enable_temporal_mvp_flag) is 0 (S227).

[00233] If the temporal candidate usage permission flag is 0 (S at S227), the spatial scale scaling prediction vector prediction candidate derivation unit 151 will derive zero or one spatial vector prediction candidate from the LX forecast (S228) and terminate the process. The detailed operation of the spatial scale scaling prediction vector prediction candidate derivation unit 151 will be described later.

[00234] If the temporal candidate use permission flag is 1 (N in S227), the spatial scale scaling prediction vector prediction candidate derivation unit 151 will skip step S228 and terminate the process.

[00235] As described above, zero or one spatial vector prediction candidates are derived from each of the groups and zero to two spatial vector prediction candidates are derived for the LX prediction.

DETAILED OPERATION OF THE SPATIAL SCALE SCALING PREDICTION VECTOR PREDICTION CANDIDATE DERIVATION UNIT 151

[00236] An explanation is now given on the detailed operation of the spatial scale scaling prediction vector prediction candidate derivation unit 151. Fig. 29 is a flowchart explaining an operation of a prediction candidate derivation unit Spatial Scale Scaling Prediction Vector Method 151. An explanation is given regarding deriving a spatial vector prediction candidate from LX prediction.

[00237] The spatial scale scaling prediction vector prediction candidate derivation unit 151 repeats the following processes for a candidate block in the i-th group (i is 1 or 2) (S230 to S236). The first group is scanned from E block to A block in order and the second group is scanned from C block to B block to D block in order.

[00238] The following processes are repeated for each candidate block in order of prediction L0 and prediction L1 (S231 to S235). Henceforth, an explanation is given while specifying the L0 forecast and the L1 forecast as the LN forecast.

[00239] Subsequently, the spatial scale scaling prediction vector prediction candidate derivation unit 151 checks whether the prediction LN of the candidate block is valid (S232). The LN prediction of a candidate block being valid means that a reference index of the LN prediction of the candidate block is greater than or equal to 0.

[00240] If the LN prediction of the candidate block is valid (S at S232), the spatial scale scaling prediction vector prediction candidate derivation unit 151 checks whether a reference representation indicated by the prediction reference index LN of the candidate block is not the same as a reference representation indicated by the prediction reference index LX (S233).

[00241] If the reference representations are not the same (S at S233), the spatial scale scaling prediction vector prediction candidate derivation unit 151 will derive a spatial vector prediction candidate while using a space vector of movement of the LN prediction of the candidate block as a reference vector of the LX prediction (S234) and will terminate the process. The detailed method of deriving a spatial vector prediction candidate by the spatial scale scaling prediction vector prediction candidate derivation unit 151 will be described later.

[00242] If the LN prediction of the candidate block is not valid (N in S232) or if the reference representations are not the same (N in S233), the scaled scaling prediction vector prediction candidate derivation unit Spatial 151 checks L1 prediction if LN prediction is L0 prediction and checks a subsequent candidate block if LN prediction is L1 prediction (S235).

[00243] Spatial scale scaling prediction vector prediction candidate derivation unit 151 ends the process when verification is finished for all candidate blocks (S236).

[00244] An explanation is now given regarding the detailed method of deriving a spatial vector prediction candidate by the spatial scale scaling prediction vector prediction candidate derivation unit 151. Using a distance of inter- representation between a reference representation NeiRefLXPic referred to by the prediction reference vector LX and a target representation CurPic, an inter-representation distance between a reference image RefLXPic indicated by the prediction reference index LX and the target representation CurPic and the vector of LX prediction reference movement as td, tb and mvLX, respectively, a spatial vector prediction candidate mvScLX is calculated by Expression 1 or Expression 2 through Expression 4 described above.

DETAILED TIME VECTOR PREDICTION CANDIDATE DERIVATION UNIT OPERATION 152

[00245] An explanation of the detailed operation of the time vector predictor candidate derivation unit 152 is now given. Fig. 30 is a flowchart explaining the operation of the time vector predictor candidate derivation unit 152. an explanation regarding the derivation of a time vector prediction candidate from the LX prediction.

[00246] The following processes are repeated for the temporal candidate blocks (S240 to S243). Temporal candidate blocks are checked in order from block H to block I.

[00247] First, the time vector prediction candidate derivation unit 152 checks whether the prediction LN of the candidate block is valid (S241). In this case, assuming that the LN forecast is the same as the LX forecast. However, the LN forecast need not be the same as the LX forecast. The LN prediction of a candidate block being valid means that a reference index of the LN prediction of the candidate block is greater than or equal to 0.

[00248] If the candidate block prediction LN is valid (S at S241), the time vector prediction candidate derivation unit 152 will derive a time vector prediction candidate while using a motion vector from the block prediction LN candidate as an LX prediction reference vector (S242) and terminate the process. A description of a detailed method of deriving a temporal vector prediction candidate will be described later.

[00249] If the candidate block prediction LN is not valid (N at S241), the time vector prediction candidate derivation unit 152 will check a subsequent candidate block (S243).

[00250] The time vector prediction candidate derivation unit 152 ends the process when verification is finished for all candidate blocks (S243).

[00251] An explanation is now given regarding a detailed method of deriving a temporal vector prediction candidate. Using an inter-representation distance between a ColPic that has a temporal candidate block and a ColRefLXPic, which is a representation referred to by the temporal candidate block in the LN prediction motion compensation prediction, an inter-representation distance between a picture reference RefLXPic indicated by the prediction reference index LX and a target representation CurPic and the reference motion vector of the prediction LX as td, tb and mvLX, respectively, a temporal vector prediction candidate mvLXCol is calculated by Expression 1 or by Expression 2 to Expression 4 described above.

MOTION REPRESENTATION DECODING DEVICE CONFIGURATION 200

[00252] An explanation of a moving representation decoding device according to the first embodiment is now given. Fig. 31 is a diagram showing the configuration of the representation decoding device 200 in motion according to the first embodiment. The motion representation decoding device 200 is a device that generates a reproduction representation by decoding a bit stream encoded by the motion representation coding device 100. Determining a slice type, the maximum value of an index reference that can be used in a slice, partitioning a coding block, determining a jump mode, determining a predictive block size type, determining a predictive block size and a position in a coding block of a prediction block (position information of a prediction block) and determining whether a prediction coding mode is intra are determined through a decoding control unit 208 outside the representation decoding device in motion 200 and fed to the motion representation decoding device 200. A reference representation list is constructed through a reference representation list construction unit 209 out of the moving representation decoding device 200 and fed to the moving representation decoding device 200. An explanation is given regarding a case in which a predictive coding mode it is not intra. An explanation is given regarding a representation B (slice B), which corresponds to the double prediction, in the first embodiment unless otherwise specified. For a representation P (slice P), which does not correspond to the double prediction, the prediction L1 needs to be omitted.

[00253] The moving representation decoding device 200 is achieved through hardware such as, for example, an information processing device or the like provided with a CPU (Central Processing Unit), a frame memory, a hard disk and similar. Through operation of the above constituted elements, the moving representation decoding device 200 achieves the functional constituted elements explained below. The slice type, the reference representation list, the maximum value of the reference index that can be used in the slice, and the position information and prediction block size of a prediction block undergoing decoding are assumed to be shared in the motion representation decoding device 200 and are therefore not shown.

[00254] The motion representation decoding device 200 according to the first embodiment includes a bitstream analysis unit 201, a prediction error decoding unit 202, an addition unit 203, a motion information 204, a motion compensation unit 205, a frame memory 206 and a motion information memory 207.

OPERATION OF THE MOTION REPRESENTATION DECODING DEVICE 200

[00255] An explanation is given below regarding the function and operation of each component. The bitstream analysis unit 201 analyzes a bitstream supplied by the terminal 30 in order to submit the prediction error coding data, a confluence flag, a confluence index, a prediction direction (interprediction type) of motion compensation prediction, a reference index, a vector difference and a vector prediction index for entropy decoding according to syntax. Entropy decoding is performed using a method including variable length coding such as arithmetic coding, Huffman coding or the like. The bitstream analysis unit 201 supplies the prediction error encoding data to the prediction error decoding unit 202 and supplies the confluence flag, the confluence index, the interprediction type, the reference index, the vector difference and vector prediction index to motion information reproduction unit 204.

[00256] The bitstream analysis unit 201 decodes the partition information for the coding block, the type of prediction block size, the position in the coding block of the prediction block and the used prediction coding mode in the motion representation decoding device 200 from the bit stream together with an SPS (Sequence Parameter Set) that defines a group of parameters for determining the properties of the bit stream, a PPS (Sequence Parameter Set) (Picture Parameter Set) that defines a group of parameters for determining the properties of the representation, where a portion header defines a group of parameters for determining the properties of the slice and the like.

[00257] The movement information reproduction unit 204 reproduces the movement information from the confluence signaling, the confluence index, the interprediction type, the reference index, the vector difference and the vector prediction index supplied by the bit stream analysis unit 201 and the group of candidate blocks supplied to the motion information memory 207 and supplies the motion information to the motion compensation unit 205 and the motion information memory 207. motion information reproduction unit 204 will be described later.

[00258] The motion compensation unit 205 performs motion compensation on a reference representation that is indicated by a reference index in the frame memory 206 based on the motion information supplied by the motion information reproduction unit 204 in order to generate a prediction signal. If the prediction direction is double prediction, the motion compensation unit 205 will average the respective prediction signals for the prediction L0 and the prediction L1 as the prediction signal and supply the prediction signal to the addition unit 203.

[00259] The prediction error decoding unit 202 generates a prediction error signal by performing a process such as, for example, inverse quantization, inverse orthogonal transformation or the like on the prediction error encoding data supplied by the analysis unit bit stream 201 and supplies the prediction error signal to the addition unit 203.

[00260] The addition unit 203 adds the prediction error signal supplied by the prediction error decoding unit 202 and the prediction signal supplied by the motion compensation unit 205 so as to generate a decoding representation signal and supply the decode representation signal to frame memory 206 and terminal 31.

[00261] The frame memory 206 and the motion information memory 207 have the same respective functions as the frame memory 110 and the motion information memory 111 of the motion representation coding device 100, respectively. The frame memory 206 stores the decoding representation signal supplied by the addition unit 203. The motion information memory 207 stores the motion information supplied by the motion information reproduction unit 204 in the units of the minimum prediction block sizes . The reference representation list building unit 209 builds a reference representation list according to the slice type supplied from the decoding control unit 208. If the slice type is a P slice or a B slice, the unit 209 will build a reference representation list L0. If slice type is B slice, reference representation list building unit 209 will build reference representation list L1. A specific method of construction of reference representation list L0 and reference representation list L1 is the same as the method by reference representation list construction unit 113 of moving representation coding device 100.

DETAILED CONFIGURATION OF THE MOTION INFORMATION REPRODUCTION UNIT 204

[00262] An explanation is now given regarding the detailed configuration of the motion information reproduction unit 204. Figure 32 shows the configuration of the motion information reproduction unit 204. The motion information reproduction unit 204 includes a coding mode decision unit 210, a motion vector reproduction unit 211 and a confluent motion information reproduction unit 212. A terminal 32, a terminal 33, a terminal 34 and a terminal 36 are connected to the coding unit 210. bitstream analysis 201, the motion information memory 207, the motion compensation unit 205 and the motion information memory 207, respectively.

DETAILED OPERATION OF MOTION INFORMATION REPRODUCTION UNIT 204

[00263] An explanation is given below regarding the function and operation of each component. The coding mode decision unit 210 determines whether the confluence signaling supplied by the bitstream analysis unit 201 is "0" or "1". If the confluence flag is "0", the coding mode decision unit 210 supplies the interprediction type, reference index, vector difference and vector prediction index supplied by the bitstream analysis unit 201 to the motion vector reproduction unit 211. If the confluence flag is "1", the coding mode decision unit 210 will supply the confluence index supplied by the bitstream analysis unit 201 to the confluent motion information 212. In the case of hopping mode, the coding mode decision unit 210 also supplies the confluence index supplied by the bitstream analysis unit 201 to the confluent motion information reproduction unit 212.

[00264] The motion vector reproduction unit 211 reproduces a motion vector from the interprediction type, the reference index, the vector difference and the vector prediction index supplied by the coding mode decision unit 210 and by the group of candidate blocks supplied by terminal 22 in order to generate the movement information and supply the movement information to terminal 34 and terminal 36.

[00265] The confluent movement information reproduction unit 212 builds a confluent movement information candidate list from the group of candidate blocks supplied by the terminal 33, selects the movement information of a confluent movement information candidate indicated by the confluence index supplied by the coding mode decision unit 210 from the confluent movement information candidate list and supplies the movement information to terminal 34 and terminal 36.

DETAILED CONFIGURATION OF THE CONFLUENT MOTION INFORMATION REPRODUCTION UNIT 212

[00266] An explanation is now given regarding the detailed configuration of the confluent motion information reproduction unit 212. Figure 33 shows the configuration of the confluent motion information reproduction unit 212. The confluent motion information reproduction unit 212 includes a confluent motion information candidate list construction unit 230 and a confluent motion information selection unit 231. A terminal 35 is connected to the coding mode decision unit 210.

DETAILED OPERATION OF CONFLUENT MOTION INFORMATION REPRODUCTION UNIT 212

[00267] An explanation is given below regarding the function and operation of each component. Fig. 34 is a diagram explaining the operation of the confluent motion information reproduction unit 212. The confluent motion information candidate list construction unit 230 builds a confluent motion information candidate list from the group of candidate blocks supplied by terminal 33 (S310) and supplies the confluent movement information candidate list to the confluent movement information selection unit 231.

[00268] The confluent movement information selection unit 231 derives a confluence index supplied by terminal 35 (S311), selects a confluent movement information candidate indicated by the confluence index from the list of confluent movement information candidates supplied by the confluent movement information candidate list construction unit 230 so as to determine confluent movement information (S312) and supply the movement information of the confluent movement information to terminals 34 and 36.

DETAILED CONFIGURATION OF MOTION VECTOR REPRODUCTION UNIT 211

[00269] Subsequently, an explanation is given regarding the detailed configuration of the motion vector reproduction unit 211. Fig. 35 is a diagram explaining the configuration of the motion vector reproduction unit 211. The vector reproduction unit The motion control unit 211 includes a vector prediction candidate list building unit 220, a vector prediction selection unit 221 and an addition unit 222. The terminal 35 is connected to the coding mode decision unit 210.

DETAILED OPERATION OF MOTION VECTOR REPRODUCTION UNIT 211

[00270] An explanation is given below regarding the function and operation of each component. Fig. 36 is a diagram explaining the operation of the motion vector reproduction unit 211. The motion vector reproduction unit 211 calculates a motion vector for prediction L0 if the type of interprediction supplied by terminal 35 is prediction L0, calculates a motion vector for the L1 forecast if the interforecast type is the L1 forecast, and calculates a motion vector for the L0 forecast and the L1 forecast if the interforecast type is the BI forecast. The calculation of a motion vector for each LX forecast is as shown in the following.

[00271] The vector prediction candidate list construction unit 220 builds a list of vector prediction candidates from the LX prediction from a reference index of the prediction LX supplied by the terminal 35 and a group of candidate blocks supplied by the terminal 33 (S320) and supplies the list of vector prediction candidates from the prediction LX to the vector prediction selection unit 221.

[00272] The vector prediction selection unit 221 derives a vector prediction index from the LX prediction supplied by terminal 35 (S321), selects a vector prediction candidate indicated by the prediction vector index from the LX prediction from the LX forecast vector list supplied by LX forecast vector list construction unit 220 so as to determine the LX forecast vector candidate as a LX forecast vector list (S322) and adds the LX forecast vector list prediction LX and a vector difference of prediction LX supplied by terminal 35 in order to calculate a motion vector of prediction LX (S323).

[00273] Motion information is generated by combining the motion vector of the LX forecast and an interforecast type, and the motion information is supplied to terminal 34 and terminal 36.

[00274] As described above, the motion representation decoding device 200 is capable of generating a representation representation by decoding a bit stream encoded by the motion representation coding device 100. MODE TWO

[00275] An explanation is given below with respect to a second embodiment. The operation of the second confluent motion candidate information supply unit 165 is different from that of the first embodiment. An explanation is given below regarding the differences of the first modality. Fig. 37 is a flow chart explaining the operation of the second confluent movement information candidate supply unit 165 according to the second embodiment. In addition to step S187 to step S189 in place of step S184 is different from the second confluent movement information candidate supply unit 165 according to the first embodiment. An explanation is given below with respect to step S187 through step S189. In that case, an explanation is given on the condition that the amount of reference representations available in the reference representation list L0 is greater than or equal to the amount of reference representations available in the reference representation list L1.

[00276] The second confluent movement information candidate supply unit 165 checks whether i is less than or equal to the maximum value of the prediction reference index L1 (num_ref_idx_l1_active_minus1) (S187). If i is less than or equal to the maximum value of the prediction reference index L1 (num_ref_idx_l1_active_minus1) (S at S187), the second confluent motion information candidate supply unit 165 derives a second supplementary confluent motion information candidate which has a motion vector (mvL1) of (0,0) for forecast L1, a reference index (ref_idx_l1) of i for forecast L1, and an interforecast type of Pred_BI (S188) and will add the second information candidate of confluent movement information to the candidate list of confluent movement information (S185). If i is not less than or equal to the maximum value of the prediction reference index L1 (num_ref_idx_l1_active_minus1) (N at S187), the second confluent motion information candidate supply unit 165 derives a second confluent motion information candidate supplement that has a motion vector (mvL1) of (0,0) for forecast L1, a reference index (ref_idx_l1) of 0 for forecast L1, and an interforecast type of Pred_BI (S189) and will add the second candidate of confluent movement information supplements the candidate list of confluent movement information (S185).

[00277] Figures 38A and 38B are diagrams explaining an example of a relationship between a reference index of a second candidate of supplemental confluent movement information and a POC according to the second embodiment. Figs. 38A and 38B show an example in which the quantity of reference representations available in the reference representation list L0 is 4 and in which the quantity of reference representations available in the reference representation list L1 is 2. Figs. 38A shows a reference index of a second candidate of supplemental confluent movement information when there is no rearrangement between the order of an input representation and the order of encoding. Figure 38B shows a reference index of a second candidate of supplemental confluent movement information when rearranging between the order of an input representation and the coding order.

[00278] An explanation is given regarding Figure 38A. An L0 prediction reference index and an L1 prediction reference index of the 0-second supplemental confluent movement information candidate are both 0, and an L0 prediction reference representation and an L1 prediction reference representation are identical, both are reference representations that have a POC of n+7. Therefore, while the 0th second supplemental confluent motion information candidate has the motion information for the dual prediction, the 0th second supplemental confluent motion information candidate represents a prediction signal that is identical to a prediction signal. for the uni-prediction obtained by performing motion compensation prediction on a reference representation that has a POC of n+7 using a motion vector of (0,0).

[00279] A reference index of prediction L0 and a reference index of forecast L1 of the first second candidate of supplemental confluent movement information are both 1 and a reference representation of prediction L0 and a reference representation of prediction L1 are identical, both are reference representations that have a POC of n+6. Therefore, while the first second supplementary confluent motion information candidate has the motion information for the dual prediction, the first second supplementary confluent motion information candidate represents a prediction signal that is identical to a prediction signal for the dual prediction. prediction obtained by performing motion compensation prediction on a reference representation that has a POC of n+6 using a motion vector of (0,0).

[00280] In the second supplemental confluent motion information candidate, the L0 forecast reference index and the L1 forecast reference index are 2 and 0, respectively, and the L0 forecast reference representation and the L1 forecast reference representation prediction L1 are a reference representation that has a POC of n+5 and a reference representation that has a POC of n+7, respectively.

[00281] In the third second supplementary confluent movement information candidate, the L0 forecast reference index and the L1 forecast reference index are 3 and 0, respectively, and the L0 forecast reference representation and the L1 forecast reference representation prediction L1 are a reference representation that has a POC of n+4 and a reference representation that has a POC of n+7, respectively.

[00282] As described above, defining the i-second supplemental confluent motion information candidate to represent the motion information with a motion vector of (0,0) for the L0 prediction and for the L1 prediction, a reference index of i for forecast L0 and for forecast L1 and an interforecast type of Pred_BI if a reference representation in the list of reference representation L1 is present and defining the i-th second candidate of confluent movement information supplement to represent motion information with a motion vector of (0,0) for forecast L0, a reference index of i for forecast L0 and a reference index of 0 for forecast L1, and an interforecast type of Pred_BI if a reference representation in the reference representation list L1 is not present, the second candidates of supplemental confluent movement information up to the number of reference representations ia available in the L1 reference representation list are derivatives, which have the movement information which is directed to the dual forecast and which represents the forecast signals which are identical to those for the uni-forecast. This allows motion compensation prediction where only a reference representation close to a target representation, which has high correlation with the target representation in general is used. This also allows double prediction to be performed in which the closest reference representation is used in a case where the second candidates of supplementary confluent motion information greater than the quantity of the reference representations available in the list of reference representation L1 are derivatives.

[00283] In that case, for example, in a case where a reference representation that has a POC of n+7 is not identical to a target representation since there is a lock object in the reference representation in a part of a moving representation and in which a reference representation that has a POC of n+6 is identical to the target representation since there is no blocking object in the reference representation that has a POC of n+6, a highly predictive signal. accurate can be obtained using the first second candidate of supplemental confluent movement information.

[00284] An explanation is given regarding Figure 38B. An L0 prediction reference index and an L1 prediction reference index of the 0-second supplemental confluent movement information candidate are both 0, and an L0 prediction reference representation and an L1 prediction reference representation are identical, both are representations of a reference that have a POC of n+4 and a reference representation that has a POC of n+6, respectively.

[00285] A reference index of prediction L0 and a reference index of prediction L1 of the first second candidate of supplemental confluent movement information are both 1 and a reference representation of prediction L0 and a reference representation of prediction L1 are a representation representation that has a POC of n+2 and a reference representation that has a POC of n+8, respectively.

[00286] A reference index of prediction L0 and a reference index of prediction L1 of the second candidate of supplementary confluent movement information are 2 and 0, respectively, and a reference representation of prediction L0 and a reference representation of prediction L1 are identical, both are reference representations that have a POC of n+6. Therefore, while the second second supplementary confluent motion information candidate has the motion information for the dual prediction, the second second supplementary confluent motion information candidate represents a prediction signal that is identical to a prediction signal for the dual prediction. prediction obtained by performing motion compensation prediction on a reference representation that has a POC of n+6 using a motion vector of (0,0).

[00287] A reference index of prediction L0 and a reference index of prediction L1 of the third second candidate of supplementary confluent movement information are 3 and 0, respectively, and a reference representation of prediction L0 and a reference representation of prediction L1 are a reference representation having a POC of n+8 and a reference representation having a POC of n+6, respectively.

[00288] As described above, all respective reference representations of the L0 prediction and the L1 prediction can be assigned to the reference representations of the second supplemental confluent motion information candidates by defining the i-second candidate of confluent motion information supplementary to have a motion vector of (0,0) for forecast L0 and forecast L1, a reference index of i for forecast L0 and forecast L1, and an interforecast type of Pred_BI. In this case, if a first supplemental confluent motion information candidate is derived by combining a second uni-forecast supplemental confluent motion information candidate from the L0 forecast and a predetermined confluent motion information candidate in the confluent motion, a new second supplemental confluent motion information candidate will not be derived if the predetermined confluent motion information candidate is targeted to the forecast forecast. Also, even when the predetermined confluent motion information candidate is directed to the dual prediction, since there is no motion vector or reference representation for the L1 prediction to be combined with the L0 prediction of the predetermined confluent motion information candidate, a supplementary first confluent motion information candidate using prediction L0 of the predetermined confluent motion information candidate is not derived.

[00289] In this case, by allowing all second supplemental confluent motion information candidates to be directed to the dual prediction, the derivation of more first supplemental confluent motion information candidates may be possible when combined with a motion information candidate predetermined confluent in the candidate list of confluent movement information, and the coding efficiency can thereby be improved. Also, even when second supplemental confluent motion information candidates are combined with each other, the probability of a first supplemental confluent motion information candidate being derived can be increased.

THIRD MODALITY

[00290] An explanation is given below with respect to a third embodiment. The operation of the second confluent motion candidate information supply unit 165 is different from that of the first embodiment. An explanation is given below regarding the differences of the first modality. Fig. 39 is a flowchart explaining the operation of the second confluent movement information candidate supply unit 165 according to the second embodiment. In addition to step S190 to step S192 in place of step S184 is different from the second confluent movement information candidate supply unit 165 according to the first embodiment. Also, an expression for calculating MaxNumGenCand, which is the maximum amount for deriving second candidates from supplemental confluent motion information, is different. An explanation is given below with respect to step S190 through step S192.

[00291] First, the second confluent motion information candidate supply unit 165 calculates MaxNumGenCand, which is the maximum amount to derive the second supplementary confluent motion information candidates, via Expression 7 (S180). MaxNumGenCand = Min(MaxNumMergeCand - NumCandList, Min(num_ref_idx_l0_active_minus1+1, num_ref_idx_l1_active_minus1+1) Expression 7

[00292] The second confluent movement information candidate supply unit 165 checks whether there is a reference representation indicated by a prediction reference index L0 in a prediction reference representation list L1 (S190).

[00293] If there is the reference representation indicated by the prediction reference index L0 in the prediction reference representation list L1, the second confluent movement information candidate supply unit 165 will derive a second supplementary confluent movement information candidate which has a motion vector (mvL1) of (0,0) for forecast L1, a forecast reference index L1 that indicates a reference representation identical to the reference representation indicated by forecast reference index L0, and a type of Pred_BI interforecast (S191) and add the second supplemental confluent movement information candidate to the confluent movement information candidate list (S185). In this case, assuming that getRefIdxL1(ref_idx_l0) is a function to obtain the reference index of prediction L1 that indicates the reference representation identical to the reference representation indicated by the reference index of prediction L0.

[00294] If there is no reference representation indicated by the prediction reference index L0 in the prediction reference representation list L1, the second confluent movement information candidate supply unit 165 will derive a second confluent movement information candidate supplement that has a motion vector (mvL1) of (0,0) for the L1 forecast, a reference index of i for the L1 forecast, and an interforecast type of Pred_BI (S192) and will add the second candidate of motion information confluent supplement to the confluent movement information candidate list (S185).

[00295] In the case where there is no rearrangement between the order of an input representation and the coding order and the reference representation list of prediction L0 is known to be identical to that of prediction L1, step S190 and the step S191 can also be set not to be performed.

[00296] Figures 40A and 40B are diagrams explaining an example of a relationship between a reference index of a second candidate of supplemental confluent movement information and a POC according to the third embodiment. Figs. 40A and 40B show an example in which the quantity of reference representations available in the reference representation list L0 is 4 and in which the quantity of reference representations available in the reference representation list L1 is 4. Figs. 40A shows a reference index of a second candidate of supplemental confluent movement information when there is no rearrangement between the order of an input representation and the order of encoding. Figure 40B shows a reference index of a second candidate of supplemental confluent movement information when rearranging between the order of an input representation and the coding order.

[00297] An explanation is given with respect to Figure 40A. An L0 prediction reference index and an L1 prediction reference index of the 0-second supplemental confluent movement information candidate are both 0, and an L0 prediction reference representation and an L1 prediction reference representation are identical, both are reference representations that have a POC of n+7. Therefore, while the 0-th second supplemental confluent motion information candidate has the motion information for the dual prediction, a prediction representation that is obtained is identical to a prediction signal for the uni-prediction obtained by performing a prediction of motion compensation in a reference representation that has a POC of n+7 using a motion vector of (0,0).

[00298] A reference index of prediction L0 and a reference index of prediction L1 of the first second candidate of supplemental confluent movement information are both 1, and a reference representation of prediction L0 and a reference representation of prediction L1 are identical, both are the reference representations that have a POC of n+6. Therefore, while the first second candidate of supplemental confluent motion information has the motion information for the dual forecast, a forecast representation that is obtained is identical to a forecast signal for the uni-forecast obtained by performing offset forecasting. motion in a reference representation that has a POC of n+6 using a motion vector of (0,0).

[00299] A reference index of prediction L0 and a reference index of prediction L1 of the second supplemental confluent movement information candidate are both 2 and a reference representation of prediction L0 and a reference representation of prediction L1 are identical, both are reference representations that have a POC of n+5. Therefore, while the second supplemental confluent motion information candidate has the motion information for the dual prediction, a prediction representation that is obtained is identical to a prediction signal for the uni-prediction obtained by performing offset prediction. motion in a reference representation that has a POC of n+5 using a motion vector of (0,0).

[00300] A reference index of prediction L0 and a reference index of prediction L1 of the third second candidate of supplementary confluent movement information are both 3 and a reference representation of prediction L0 and a reference representation of prediction L1 are identical, both are reference representations that have a POC of n+4. Therefore, while the third and second supplemental confluent motion information candidate has the motion information for the dual prediction, a prediction representation that is obtained is identical to a prediction signal for the uni-prediction obtained by performing offset compensation prediction. motion in a reference representation that has a POC of n+4 using a motion vector of (0,0).

[00301] An explanation is given regarding Figure 40B. An L0 prediction reference index and an L1 prediction reference index of the 0-second supplemental confluent motion information candidate are 0 and 2, respectively, and an L0 prediction reference representation and an L1 prediction reference representation are identical, both are reference representations that have a POC of n+4 Therefore, while the 0-th second supplemental confluent motion information candidate has the motion information for the dual prediction, a prediction representation that is obtained is identical to a prediction signal for the uni-prediction obtained by performing motion compensation prediction on a reference representation having a POC of n+4 using a motion vector of (0,0).

[00302] An L0 prediction reference index and an L1 prediction reference index of the first second supplemental confluent motion information candidate are 1 and 3, respectively, and an L0 prediction reference representation and a prediction reference representation L1 are identical, both are the reference representations that have a POC of n+2. Therefore, while the first second candidate of supplemental confluent motion information has the motion information for the dual forecast, a forecast representation that is obtained is identical to a forecast signal for the uni-forecast obtained by performing offset forecasting. motion in a reference representation that has a POC of n+2 using a motion vector of (0,0).

[00303] An L0 prediction reference index and an L1 prediction reference index of the second supplemental confluent motion information candidate are 2 and 0, respectively, and an L0 prediction reference representation and a prediction reference representation L1 are identical, both are reference representations that have a POC of n+6. Therefore, while the second supplemental confluent motion information candidate has the motion information for the dual prediction, a prediction representation that is obtained is identical to a prediction signal for the uni-prediction obtained by performing offset prediction. motion in a reference representation that has a POC of n+6 using a motion vector of (0,0).

[00304] An L0 prediction reference index and an L1 prediction reference index of the third second supplementary confluent movement information candidate are 3 and 1, respectively, and an L0 prediction reference representation and a prediction reference representation L1 are identical, both are reference representations that have a POC of n+8. Therefore, while the third and second supplemental confluent motion information candidate has the motion information for the dual prediction, a prediction representation that is obtained is identical to a prediction signal for the uni-prediction obtained by performing offset compensation prediction. motion in a reference representation that has a POC of n+8 using a motion vector of (0,0).

[00305] As described above, if there is a reference representation indicated by a prediction reference index L0 in the prediction reference representation list L1, the second confluent movement information candidate supply unit 165 can derive a second candidate of supplemental confluent motion information having the motion information directed to the dual forecast while representing a forecast signal identical to a forecast signal for the uni-forecast by setting the L1 forecast reference index to indicate a reference representation which is identical to a reference representation indicated by the forecast reference index L0.

[00306] In this case, for example, in a case where a representation located at the same position as a target block in the nearest reference representation is not identical to a representation of the target block since there is a blocking object at the same position that the target block in the reference representation is closer and that a representation lying at the same position as the target block in a reference representation that is currently distant to an extent is identical to the representation of the target block since the target block no longer exists blocking object, a highly accurate prediction signal can be obtained using a second candidate of supplemental confluent motion information that refers to the reference representation without the blocking object. More specifically, in Figure 40A, in a case where a representation located at the same position as a target block in a reference representation that has a POC of n+7 is not identical to a representation of the target block since there is an object blocking in the reference representation that has a POC of n+7 and that a representation located at the same position as the target block in a reference representation that has a POC of n+6 is identical to the target block representation since the blocking object no longer exists in the reference representation which has a POC of n+6, a highly accurate prediction signal can be obtained by selecting the first second candidate of supplemental confluent motion information.

[00307] Also, for example, in moving representations where the representations before a target representation are stationary and where the representations after the target representation have a movement, a highly accurate prediction signal can be obtained using a representation reference that is temporarily before the target representation. More specifically, in Figure 40B, in moving representations where the representations before a target representation are stationary and where a reference representation that has a POC of n+6 and subsequent representations thereof have a movement, a sign of Highly accurate prediction can be obtained using a 0-th second candidate of supplemental confluent motion information that refers to a reference representation that has a POC of n+4, which is temporally before the target representation.

[00308] Also, by assigning all the reference representations of the L0 prediction and the L1 prediction to the reference representations of second supplementary confluent motion information candidates, all the prediction signals of the second supplementary confluent motion information candidates can be made to be.

[00309] Furthermore, since step S175 is provided to the first confluent movement information candidate supply unit 164, a supplementary first confluent movement information candidate that is identical to a second supplementary confluent movement information candidate is not added in the first confluent motion information candidate supply unit 164 and upon increasing the probability of a confluent motion information candidate having the different motion information that is present in a confluent motion information candidate list , coding efficiency can be improved.

[00310] The moving representation bitstream emitted from the moving representation encoding device according to any of the embodiments described above has a specific data format so that it can be decoded according to the method coding used in the modalities. The motion representation decoding device compatible with the motion representation coding device is capable of decoding the bit stream of the specific data format.

[00311] If a wired or wireless network is used to exchange the bitstreams between the motion representation encoding device and the motion representation decoding device, the bitstream can be converted into a data format suitable for the mode of transmission over a communication channel and be transmitted accordingly. In that case, a motion representation transmission device is provided for converting the bit streams emitted from the motion representation encoding device into encoded data of a data format suitable for the mode of transmission over the communication channel and for transmitting the bit streams over the network and a motion representation receiving device for receiving the encoded data over the network for recovering the bit streams and supplying the recovered bit streams to the motion representation decoding device.

[00312] The motion representation transmission device includes a memory for buffering bit streams emitted from the motion representation coding device, a packet processing unit for packaging the bit streams, and a transmission unit for transmit the packetized encoding data over the network. The motion representation receiving device includes a retrieval unit for receiving the packetized encoded data over the network, a memory for buffering the received encoded data, and a packet processing unit for subjecting the encoded data to an unpackaging process in order to generating the bitstreams and supplying the generated bitstreams to the motion representation decoding device.

[00313] The processes described above relating to encoding and decoding can of course be implemented via a hardware-based apparatus for transmission, storage or reception. Alternatively, processes can be deployed via firmware stored in read-only memory (ROM), flash memory, etc., or via software on a computer, etc. The firmware program or software program may be available, for example, on a computer-readable recording medium. Alternatively, programs can be made available from a server over a wired or wireless network. Yet alternatively, the programs may be made available in the form of data transmission via digital terrestrial or satellite broadcasting systems.

[00314] An explanation of the present invention based on the embodiments is described above. The embodiment is intended to be illustrative only and it will be obvious to those skilled in the art that various modifications to the constituent elements and processes can be developed and that such modifications are also within the scope of the present invention. DESCRIPTION OF REFERENCE NUMBERS 100 - Moving representation encoding device, 101 - Prediction block representation derivation unit, 102 - Subtraction unit, 103 - Prediction error coding unit, 104 - Current generation unit bit length, 105 - prediction error decoding unit, 106 - motion compensation unit, 107 - addition unit, 108 - motion vector detection unit, 109 - motion information generation unit, 110 - memory 111 - motion information memory, 112 - coding control unit, 113 - reference representation list building unit, 120 - vector prediction mode determination unit, 121 - mode determination unit of confluence, 122 - prediction coding mode determination unit, 130 - vector prediction candidate list construction unit, 131 - vector prediction determination unit, 140 - confluent motion information candidate list construction unit, 141 - confluent motion information candidate selection unit, 150 - spatial vector predictor candidate derivation unit, 151 - spatial vector prediction candidate derivation unit spatial scale scaling prediction, 152 - temporal vector prediction candidate derivation unit, 153 - vector prediction list building unit, 154 - redundant vector prediction candidate deletion unit, 155 - supplying unit of vector prediction candidate, 160 - spatial confluent motion information candidate derivation unit, 161 - temporal confluent motion information candidate derivation unit, 162 - confluent motion information candidate list construction unit, 163 - redundant confluent movement information candidate deletion unit, 164 - first supply unit of confluent motion information candidate, 165 - second confluent motion information candidate supply unit, 200 - motion representation decoding device, 201 - bit stream analysis unit, 202 - motion error decoding unit prediction, 203 - adding unit, 204 - motion information reproduction unit, 205 - motion compensation unit, 206 - frame memory, 207 - motion information memory, 208 - decoding control unit, 209 - reference representation list construction unit, 210 - coding mode decision unit, 211 - motion vector reproduction unit, 212 - confluent movement information reproduction unit, 220 - candidate list construction unit of vector prediction, 221 - vector prediction selection unit, 222 - addition unit, 230 - movement information candidate list construction unit confluent movement, 231 - confluent movement information selection unit INDUSTRIAL APPLICABILITY

[00315] The present invention is applicable to motion encoding and decoding technology for encoding and decoding motion information used in motion compensation prediction.

Claims (4)

1. Motion representation decoding device characterized in that it is adapted to derive motion information including a reference index and a motion vector in units of prediction blocks, comprising: a decoding unit configured to decode, from from a bit stream, a specification index for specifying a confluent motion information candidate used for a prediction block undergoing decoding; a spatial confluent motion information candidate derivation unit configured to derive a spatial confluent motion information candidate from motion information of a plurality of decoded prediction blocks adjacent to the prediction block undergoing decoding; a temporal confluent motion information candidate derivation unit configured to derive a temporal confluent motion information candidate from motion information of a prediction block into a decoded representation other than a representation in which the prediction block undergoing decoding is placed on; a confluent motion information candidate list construction unit configured to generate a confluent motion information candidate list, which is a confluent motion information candidate list, using the spatial confluent motion information candidate and the candidate temporal confluent movement information; a confluent motion information candidate supply unit configured to generate a new confluent motion information candidate including, if the reference index indicates an available reference representation, the reference index and, if the reference index does not indicate a available reference representation, a predetermined reference index together with a motion vector having a size and direction that are preset and for adding the new confluent motion information candidate to the list of confluent motion information candidates; and a confluent movement information selection unit configured to select a confluent movement information candidate from the list of confluent movement information candidates to which the new confluent movement information candidate has been added and to derive the movement information candidate confluent selected as the motion information of the prediction block subjected to decoding, based on the specification index that was decoded, where the predetermined reference index is 0. 2. Motion representation decoding device characterized in that it is adapted to derive motion information including a reference index and a motion vector in units of prediction blocks, comprising: a decoding unit configured to decode, from from a bit stream, a specification index for specifying a confluent motion information candidate used for a prediction block undergoing decoding; a spatial confluent motion information candidate derivation unit configured to derive a spatial confluent motion information candidate from motion information of a plurality of decoded prediction blocks adjacent to the prediction block undergoing decoding; a temporal confluent motion information candidate derivation unit configured to derive a temporal confluent motion information candidate from motion information of a prediction block into a decoded representation other than a representation in which the prediction block undergoing decoding is placed on; a confluent motion information candidate list construction unit configured to generate a confluent motion information candidate list, which is a confluent motion information candidate list, using the spatial confluent motion information candidate and the candidate temporal confluent movement information; a confluent motion information candidate supply unit configured to generate a new confluent motion information candidate including, if the reference index indicates an available reference representation, the reference index and, if the reference index does not indicate a available reference representation, a predetermined reference index together with a motion vector having a size and direction that are preset and for adding the new confluent motion information candidate to the list of confluent motion information candidates; and a confluent movement information selection unit configured to select a confluent movement information candidate from the list of confluent movement information candidates to which the new confluent movement information candidate has been added and to derive the movement information candidate confluent selected as the motion information of the prediction block subjected to decoding, based on the specification index that has been decoded, wherein the motion vector having a size that is preset is (0,0). 3. Motion representation decoding device characterized in that it is adapted to derive motion information including a reference index and a motion vector in units of prediction blocks, comprising: a decoding unit configured to decode, from from a bit stream, a specification index for specifying a confluent motion information candidate used for a prediction block undergoing decoding; a spatial confluent motion information candidate derivation unit configured to derive a spatial confluent motion information candidate from motion information of a plurality of decoded prediction blocks adjacent to the prediction block undergoing decoding; a temporal confluent motion information candidate derivation unit configured to derive a temporal confluent motion information candidate from motion information of a prediction block into a decoded representation other than a representation in which the prediction block undergoing decoding is placed on; a confluent motion information candidate list construction unit configured to generate a confluent motion information candidate list, which is a confluent motion information candidate list, using the spatial confluent motion information candidate and the candidate temporal confluent movement information; a confluent motion information candidate supply unit configured to generate a new confluent motion information candidate including, if the reference index indicates an available reference representation, the reference index and, if the reference index does not indicate a available reference representation, a predetermined reference index together with a motion vector having a size and direction that are preset and for adding the new confluent motion information candidate to the list of confluent motion information candidates; and a confluent movement information selection unit configured to select a confluent movement information candidate from the list of confluent movement information candidates to which the new confluent movement information candidate has been added and to derive the movement information candidate confluent selected as the motion information of the prediction block undergoing decoding, based on the specification index that has been decoded, wherein the confluent motion information candidate supply unit sequentially generates a new confluent motion information candidate having, as the reference index, a value incremented from 0 by one at a time and assigns the predetermined reference index in a case where the value incremented by one at a time exceeds the maximum value of an available reference index. 4. Motion representation decoding device characterized in that it is adapted to derive motion information including a reference index and a motion vector in units of prediction blocks, comprising: a decoding unit configured to decode, from from a bit stream, a specification index for specifying a confluent motion information candidate used for a prediction block undergoing decoding; a spatial confluent motion information candidate derivation unit configured to derive a spatial confluent motion information candidate from motion information of a plurality of decoded prediction blocks adjacent to the prediction block undergoing decoding; a temporal confluent motion information candidate derivation unit configured to derive a temporal confluent motion information candidate from motion information of a prediction block into a decoded representation other than a representation in which the prediction block undergoing decoding is placed on; a confluent motion information candidate list construction unit configured to generate a confluent motion information candidate list, which is a confluent motion information candidate list, using the spatial confluent motion information candidate and the candidate of temporal confluent movement information; a first confluent motion information candidate supply unit configured to generate a new confluent motion information candidate obtained by combining a first confluent motion information candidate and a second confluent motion information candidate included in the information candidate list of confluent motion information and add the new confluent motion information candidate to the list of confluent motion information candidates; a second confluent motion information candidate supply unit configured to generate a new confluent motion information candidate including, if the reference index indicates an available reference representation, the reference index and, if the reference index does not indicate an available reference representation, a predetermined reference index together with a motion vector that has a size and direction that are predefined, and add the new confluent motion information candidate to the list of confluent motion information candidates to which the candidate of confluent movement information was added by the first candidate supply unit of confluent movement information; and a confluent movement information selection unit configured to select a confluent movement information candidate from the list of confluent movement information candidates to which the confluent movement information candidate has been added by the second confluent movement information candidate supply unit confluent motion information and derive the selected confluent motion information candidate as the motion information of the prediction block undergoing decoding, based on the specification index that was decoded.

Images