JP6638366B2 - Video coding apparatus, video coding method, and video coding program - Google Patents

Description

  The present invention relates to a video coding technique, and more particularly to a video coding technique based on motion vector merging.

  Patent Literature 1 describes an example of a technique for encoding a video. The video coding apparatus described in Patent Literature 1 codes a video block using one of bidirectional prediction on two reference pictures and unidirectional prediction on one reference picture.

JP 2014-506084 A

  In the technique of Patent Document 1, when there is one reference picture, only unidirectional prediction is used. Therefore, prediction cannot always be performed with high accuracy.

  An object of the present invention is to provide a video encoding device and the like that can improve the accuracy of prediction even when there is one reference picture.

  A video encoding device according to an aspect of the present invention includes a video encoding device that includes a code included in a pre-refinement vector selected as a reference for searching for a predicted motion vector that indicates prediction of motion of a block into which a picture included in an encoding target video is divided. The number of candidate interpolation vectors generated as candidate vectors including the pre-refinement vector of the block adjacent to the current block and the number of inherited motion vectors obtained as the pre-refinement vector of the current block are 1. And a motion vector compensation unit for selecting the compensation vector, which is compensated as the pre-refinement vector, from the candidate vectors.

  The video encoding method according to an aspect of the present invention provides a video encoding method that includes, among the pre-refinement vectors selected as a reference for searching for a predicted motion vector that indicates prediction of motion of a block into which a picture included in a video to be encoded is divided, Generating a candidate vector including the unrefined vector of the block adjacent to the block to be encoded, and performing the pre-refining if the number of inherited motion vectors obtained as the unrefined vector of the encoding target block is 1 The compensation vector to be supplemented as a vector is selected from the candidate vectors.

  A video encoding program according to an aspect of the present invention provides a computer with a pre-refining vector selected as a reference for searching for a predicted motion vector that represents prediction of motion of a block into which a picture included in an encoding target video is divided. Among them, the interpolation candidate generation processing for generating a candidate vector including the unrefined vector of the block adjacent to the encoding target block, and the number of inherited motion vectors obtained as the unrefined vector of the encoding target block are A motion vector compensation process of selecting, from the candidate vectors, the compensation vector compensated as the pre-refinement vector when the value is 1.

  The present invention has an effect that the accuracy of prediction can be improved even when there is one reference picture.

FIG. 1 is a block diagram illustrating an example of an overall configuration of a video encoding device according to a comparative example of the present invention. FIG. 2 is a block diagram schematically illustrating an example of a configuration of a motion compensation unit of a video encoding device according to a comparative example of the present invention. FIG. 3 is a block diagram illustrating an example of a configuration of a motion compensation unit of a video encoding device according to a comparative example of the present invention. FIG. 4 is a block diagram schematically illustrating an example of a configuration of a motion vector compensation unit according to a comparative example of the present invention. FIG. 5 is a diagram schematically illustrating positions of blocks from which a motion vector is selected by the compensation candidate generation unit according to the comparative example and the first embodiment of the present invention. FIG. 6 is a block diagram schematically illustrating an example of a configuration of a motion vector compensation unit according to a comparative example of the present invention. FIG. 7 is a diagram illustrating an example of a structure of a portion that performs motion vector compensation and refinement in a video encoding device according to a comparative example of the present invention. FIG. 8 is a flowchart illustrating an example of an operation of generating a predicted image of an encoding target image by the video encoding device of the comparative example of the present invention. FIG. 9 is a flowchart illustrating an example of the operation of the motion vector compensation processing of the video encoding device 1 according to the comparative example of the present invention. FIG. 10 is a block diagram illustrating an example of a configuration of the video encoding device according to the first embodiment of the present invention. FIG. 11 is a block diagram schematically illustrating an example of the configuration of the motion compensation unit 200 of the video encoding device according to the first embodiment of the present invention. FIG. 12 is a block diagram illustrating an example of a configuration of the motion compensation unit according to the first embodiment of this invention. FIG. 13 is a block diagram showing in detail an example of the configuration of the motion vector compensation unit according to the first embodiment of the present invention. FIG. 14 is a block diagram schematically illustrating an example of the configuration of the motion vector compensation unit according to the first embodiment of the present invention. FIG. 15 is a first flowchart illustrating the operation of the motion compensation unit according to the first embodiment of the present invention. FIG. 16 is a second flowchart illustrating the operation of the motion compensation unit according to the first embodiment of the present invention. FIG. 17 is a flowchart illustrating an example of an operation of a motion vector compensation process by the motion vector compensation unit according to the first embodiment of the present invention. FIG. 18 is a diagram schematically illustrating a portion that performs motion vector compensation and refinement in the video encoding device according to the first embodiment of this invention. FIG. 19 is a block diagram illustrating an example of a configuration of a video encoding device according to the second embodiment of the present invention. FIG. 20 is a flowchart illustrating an example of the operation of the video encoding device according to the present embodiment. FIG. 21 is a diagram illustrating an example of a hardware configuration of a computer capable of realizing the video encoding devices according to the embodiment and the comparative example. FIG. 22 is a block diagram illustrating an example of a configuration of a video encoding device 1 implemented by a circuit according to a comparative example of the present invention. FIG. 23 is a block diagram illustrating an example of a configuration of a motion compensation circuit 1100 according to a comparative example of the present invention. FIG. 24 is a block diagram illustrating an example of a configuration of a motion compensation circuit 1100 according to a comparative example of the present invention. FIG. 25 is a block diagram illustrating an example of a configuration of a motion vector compensation circuit 1010 according to a comparative example of the present invention. FIG. 26 is a block diagram illustrating an example of a configuration of a motion vector compensation circuit 1010 according to a comparative example of the present invention. FIG. 27 is a block diagram illustrating an example of a configuration of a video encoding device 2 implemented by a circuit according to the third embodiment of this invention. FIG. 28 is a block diagram illustrating an example of a configuration of the motion compensation circuit 1200 according to the first embodiment of this invention. FIG. 29 is a block diagram illustrating an example of a configuration of a motion compensation circuit 1200 implemented by a circuit according to the first embodiment of this invention. FIG. 30 is a block diagram illustrating an example of a configuration of the motion vector compensation circuit 1090 according to the first embodiment of this invention. FIG. 31 is a block diagram illustrating an example of a configuration of the motion vector compensation circuit 1090 according to the first embodiment of this invention. FIG. 32 is a block diagram illustrating an example of a configuration of a video encoding device 2A implemented by a circuit according to the second embodiment of this invention.

  Hereinafter, first, a comparative example which is a related art of the present invention will be described, and then an embodiment of the present invention will be described. In the block diagrams illustrating the configurations of the devices, units, circuits, and the like according to the comparative example and the embodiment of the present invention, the data flow is not limited to the direction of the arrow.

[Comparative example]
First, a comparative example as an object to be compared with the embodiment of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram illustrating an example of an overall configuration of a video encoding device according to a comparative example of the present invention. The video encoding device 1 illustrated in FIG. 1 is a device that encodes a received video. Specifically, the video encoding device 1 receives a video, and encodes the received video according to, for example, one of the encoding schemes. Then, the video encoding device 1 outputs a bitstream in which the video is encoded. The encoding method (that is, standard) is described in, for example, 265 / HEVC (High Efficiency Video Coding). The encoding method is, for example, H.264. H.264 / MPEG-4 AVC (Moving Picture Experts Group Advanced Video Coding). In the following, the encoding method is H.264. The case of H.265 / HEVC will be described.

  Referring to FIG. 1, a video encoding device 1 of the comparative example includes a block division unit 101, a pre-analysis unit 102, an intra prediction unit 103, a switching unit 104, an integer conversion unit 105, and a quantization unit 106 , An inverse quantization unit 107, an inverse transformation unit 108, and an encoding unit 109. The video encoding device 1 further includes a filter unit 116, a frame storage unit 117, a subtraction unit 118, and an addition unit 119. The video encoding device 1 of the present comparative example further includes a motion compensation unit 100.

  The block dividing unit 101 divides a picture of a digitized video into blocks. The intra prediction unit 103 performs intra prediction. The switching unit 104 switches between an output from the motion compensation unit 100 described later and an output from the intra prediction unit 103. The subtraction unit 118 calculates a difference between the input video and the output from the motion compensation unit 100 or the output from the intra prediction unit 103 selected by switching by the switching unit 104. The integer conversion unit 105 performs integer conversion. The quantization unit 106 performs quantization. The encoding unit 109 performs the encoding and outputs a bit stream obtained by the encoding. The inverse quantization unit 107 performs inverse quantization. The inverse integer conversion unit 108 performs an inverse integer conversion. The addition unit 119 adds the output of the inverse integer conversion unit 108 and the output from the motion compensation unit 100 or the output from the intra prediction unit 103 selected by the switching by the switching unit 104. The filter unit 116 performs filtering using a deblocking filter or the like. The frame buffer 217 stores encoded pictures. Each of the above elements performs an operation according to the above standard. Therefore, a detailed description of each of the above elements will be omitted.

  The pre-analysis unit 102 detects a motion vector in an input picture (that is, a current picture to be coded) using, for example, a coded picture stored in the frame buffer 217. For example, when the current picture is a P-picture, the pre-analysis unit 102 transmits one motion vector to the motion compensation unit 100. When the current picture is a B picture, the pre-analysis unit 102 transmits one or two motion vectors to the motion compensation unit 100. When the current picture is a B picture, the pre-analysis unit 102 derives, for example, two motion vectors. If any of the motion vectors is determined to be optimal based on a predetermined determination criterion, the pre-analysis unit 102 may transmit only the motion vector determined to be optimal to the motion compensation unit 100. . In that case, the pre-analysis unit 102 transmits one motion vector to the motion compensation unit 100. When there is no motion vector determined to be optimal by the preliminary analysis unit 102, the preliminary analysis unit 102 transmits two motion vectors to the motion compensation unit 100. The pre-analysis unit 102 may transmit a signal PicType indicating the type of the encoding target picture to the motion compensation unit 100.

  The reference image for the two motion vectors may be a combination of a picture after the current picture and a previous picture, or a combination of two pictures before the current picture. A combination of two later pictures may be used. In the following description, a motion vector transmitted from the pre-analysis unit 102 to the motion compensation unit 100 is referred to as a “succeeding motion vector”.

Two communication paths capable of separately transmitting two inherited motion vectors may be provided between the pre-analysis unit 102 and the motion compensation unit 100. As described later, the pre-analysis unit 102 and the motion compensation unit 100 may be implemented as a circuit. In that case, two signal lines capable of separately transmitting two inherited motion vectors may be provided between the pre-analysis unit 102 and the motion compensation unit 100 as the two communication paths described above. The two inherited motion vectors may be transmitted to the motion compensation unit 100 via different signal lines. When transmitting one inherited motion vector to the motion compensating unit 100, the pre-analysis unit 102 may transmit the inherited motion vector using one of the signal lines. Then, the pre-analysis unit 102 may transmit, to the motion compensation unit 100, a signal indicating a system of a signal line used for transmitting the inherited motion vector. For example, two signal lines are referred to as a first signal line and a second signal line. In the following description, the inherited motion vector transmitted via the first signal line is referred to as “mMV ₀ ”. Similarly, the inherited motion vector transmitted via the second signal line is denoted as “mMV ₁ ”.

For example, when the pre-analysis unit 102 transmits the mMV ₀ and does not transmit the mMV ₁ , the pre-analysis unit 102 transmits the mMV ₀ and transmits a signal indicating that the mMV ₁ is not transmitted to the motion compensation unit 100. For example, when transmitting the mMV ₁ without transmitting the mMV ₀ , the pre-analysis unit 102 transmits to the motion compensation unit 100 a signal indicating that the mMV ₁ is transmitted without transmitting the mMV ₀ . For example, when transmitting the mMV ₀ and the mMV ₁ , the pre-analysis unit 102 transmits a signal indicating that the mMV ₀ and the mMV ₁ are transmitted to the motion compensation unit 100. In the following description, the signal indicating the inherited motion vector to be transmitted is referred to as “mDir” (or the signal “mDir”). Further, the signal mDir is also referred to as a “prediction direction”. In the description of the comparative example of the present invention and each of the following embodiments, the value of the signal mDir indicating that the mMV ₀ is transmitted and the mMV ₁ is not transmitted is also referred to as “L0”. The value of the signal mDir indicating that the mMV ₀ is not transmitted and the mMV ₁ is transmitted is also expressed as “L1”. Furthermore, the value of the signal mDir indicating that the mMV ₀ and the mMV ₁ are transmitted is also referred to as “Bi”.

In general, L0 and L1 each point to two reference picture lists in a B picture. The above-mentioned inherited motion vector mMV ₀ may be, for example, a motion vector in the reference image of the reference image list L0. Inheritance motion vector MMV ₁ may be, for example, a motion vector in the reference image of the reference image list L1.

  FIG. 2 is a block diagram schematically illustrating an example of the configuration of the motion compensation unit 100 of the video encoding device 1 of the comparative example. Referring to FIG. 2, the motion compensation unit 100 includes a motion vector compensation unit 10, an object selection unit 20, a refinement unit 30, a bi-prediction generation unit 40, a prediction selection unit 50, a compensation candidate generation unit 60, And a mode determining unit 70. Although not explicitly shown in FIG. 2, the refiner 30 can read an encoded picture stored in the frame storage 117 shown in FIG. 1. The motion vector transmitted from the pre-analysis unit 102 is input to the motion vector compensation unit 10 and the target selection unit 20. The picture type PicType transmitted from the pre-analysis unit 102 is input to the motion vector compensation unit 10.

  The compensation candidate generation unit 60 generates a predicted motion vector of a block around the current block and a predicted motion vector of a picture immediately before the picture including the current block. The compensation candidate generation unit 60 transmits the generated motion vector to the motion vector compensation unit 10.

  The mode determination unit 70 transmits a signal indicating a compensation mode described later to the motion vector compensation unit 10. The compensation mode may be determined experimentally in advance. In this case, the mode determination unit 70 may transmit a signal indicating a compensation mode that has been experimentally determined in advance to the motion vector compensation unit 10. The compensation mode may be determined in advance for each picture type. In this case, the mode determination unit 70 may transmit a signal indicating the compensation mode determined for the type of the picture including the current block to the motion vector compensation unit 10. The compensation mode may be dynamically determined during execution of the encoding process. In this case, for example, the mode determination unit 70 may evaluate a coding error by comparing, for example, a picture with an image in which the picture is coded and decoded. Then, the mode determination unit 70 may use the result of the error evaluation to dynamically select the compensation mode so as to reduce the error. Note that FIG. 2 does not illustrate the flow of data when the mode determination unit 70 compares a picture with a composite image.

The motion vector compensation unit 10 selects a compensation motion vector from the motion vectors received from the compensation candidate generation unit 60 according to the selection method in the compensation mode received. The compensation mode and its selection method will be described later in detail. The motion vector compensation unit 10 may select the compensation motion vector when either one of the inherited motion vector mMV ₀ and the inherited motion vector mMV ₀ is not input.

When the inherited motion vector mMV ₀ is not input, the target selection unit 20 inputs a supplementary motion vector to the refine unit 30 instead of the inherited motion vector mMV ₀ . When the inherited motion vector mMV ₁ is not input, the target selection unit 20 inputs a supplementary motion vector to the refine unit 30 instead of the inherited motion vector mMV ₁ .

  The refiner 30 detects an area (that is, a predicted image) where the encoding cost of the encoding target block is minimum in a predetermined range based on the motion vector of the reference image referenced by the input motion vector. The refiner 30 outputs, as a prediction vector, a vector from the coordinates indicating the position of the current block to the coordinates indicating the position of an area where the coding cost of the current block is minimum. The refiner 30 further outputs, as a prediction signal, an image in an area where the encoding cost of the encoding target block is minimum. The refiner 30 may detect, for each of the two input motion vectors, an area where the encoding cost of the encoding target block is minimum. Then, the refiner 30 may output the prediction vectors (rMV0 and rMV1) and the prediction signals (pred (rMV0) and pred (rMV1)). The refiner 30 may output the prediction vectors (rMV0 and rMV1) to the bi-prediction generator 40, for example. The refiner 30 may output, for example, the prediction vectors (rMV0 and rMV1) and the prediction signals (pred (rMV0) and pred (rMV1)) to the prediction selector 50.

  In this comparative example, when there are two inherited motion vectors generated by the pre-analysis unit 102, the target selecting unit 20 refines the two inherited motion vectors (that is, the inherited motion vector mMV0 and the inherited motion vector mMV1). Transmit to the unit 30. When the number of inherited motion vectors generated by the pre-analysis unit 102 is one, the target selecting unit 20 sends the supplementary motion vector and one of the inherited motion vector mMV0 and the inherited motion vector mMV1 to the refinement unit 30. Send. In the following description, the two motion vectors transmitted from the target selection unit 20 to the refinement unit 30 are also referred to as “pre-refinement motion vectors”.

  As described above, the number of inherited motion vectors is not necessarily two. However, when the number of inherited motion vectors is one, the motion vector compensation unit 10 generates a compensation motion vector. Then, the target selection unit 20 transmits the inherited motion vector and the generated supplementary motion vector to the refine unit 30 as two inherited motion vectors. The refiner 30 generates two predicted motion vectors based on the received inherited motion vectors. Therefore, the refinement unit 30 transmits two prediction vectors to the bi-prediction generation unit 40 even if the number of inherited motion vectors is one.

  The bi-prediction generation unit 40 receives the predicted motion vectors rMV0 and rMV1, and generates bi-prediction based on the received predicted motion vectors rMV0 and rMV1. The bi-prediction is an average image of the image in the area indicated by the motion vector rMV0 and the image in the area indicated by rMV1. The bi-prediction generation unit 40 outputs the generated bi-prediction signal bipred (rMV0, rMV1) representing the bi-prediction to the prediction selection unit 50.

  The prediction selecting unit 50 receives from the refiner 30 two predicted signals and a predicted motion vector indicating a predicted image represented by the predicted signal. The prediction selection unit 50 further receives a bi-prediction signal representing bi-prediction from the bi-prediction generation unit 40. As described above, even when the number of inherited motion vectors is one, the prediction selection unit 50 receives two prediction signals and a bi-prediction signal.

  The prediction selecting unit 50 selects a predicted image having the smallest encoding cost among the predicted images indicated by the two predicted signals received from the refinement unit 30 and the predicted images indicated by the bi-predicted signal received from the bi-prediction generation unit 40. Identify. The prediction selecting unit 50 outputs a predicted signal pred indicating the specified predicted image and a predicted motion vector (rMV0 or rMV1) indicating the predicted image represented by the predicted signal. When a prediction image indicating a bi-prediction signal is specified as the prediction image with the lowest encoding cost, the prediction selection unit 50 outputs two prediction motion vectors rMV0 and rMV1. The prediction selecting unit 50 further outputs a prediction direction of a predicted image represented by the prediction signal. For example, when the prediction motion vector is rMV0, the prediction selection unit 50 outputs L0 (specifically, a signal indicating a value representing L0) as the prediction direction. When the prediction motion vector is rMV1, the prediction selection unit 50 outputs L1 (specifically, a signal indicating a value representing L1) as the prediction direction. When the predicted motion vectors are rMV0 and rMV1, the prediction selecting unit 50 outputs Bi (specifically, a signal indicating a value representing Bi) as the prediction direction.

  Hereinafter, each unit included in the motion compensation unit 100 of the video encoding device 1 of the comparative example will be described in more detail.

  FIG. 3 is a block diagram illustrating an example of a configuration of the motion compensation unit 100 of the video encoding device 1 of the comparative example. FIG. 3 schematically illustrates an example in which the motion compensation unit 100 is implemented by a circuit.

  The motion compensation unit 100 of this implementation example includes a motion vector compensation unit 10, a target selection unit 20, a refinement unit 30, a bi-prediction generation unit 40, a prediction selection unit 50, a compensation candidate generation unit 60, a mode determination Unit 70. The target selection unit 20 includes a first target selection unit 21 and a second target selection unit 22. The refiner 30 includes a first refiner 31 and a second refiner 32.

  The compensation candidate generation unit 60 receives the peripheral vectors and selects a predicted motion vector from the received peripheral vectors. The peripheral vector is a predicted motion vector of a block adjacent to the current block or a predicted motion vector of a block of a picture immediately before the current block. The method in which the compensation candidate generation unit 60 selects a predicted motion vector may be a selection method in an encoding method of video encoding performed by the video encoding device 1. The method by which the compensation candidate generation unit 60 selects a predicted motion vector may be any of various other methods. The compensation candidate generation unit 60 transmits the selected predicted motion vector to the motion vector compensation unit 10.

FIG. 5 is a diagram schematically illustrating a position of a block from which a motion vector is selected by the compensation candidate generation unit 60. The compensation candidate generating unit 60 selects, for example, predicted motion vectors of any two blocks of A0, A1, B0, B1, B2, C0, and C1. In the following description, the predicted motion vectors of the two blocks C0 and C1 are particularly referred to as “temporal predicted motion vectors”. In addition, a predicted motion vector of a block of A0, A1, B0, B1, and B2 is particularly referred to as an “AMVP (Advanced Motion Vector Prediction) predicted motion vector”. The compensation candidate generation unit 60 selects two AMVP predicted motion vectors from the predicted motion vectors of the blocks A0, A1, B0, B1, and B2, for example. The compensation candidate generation unit 60 may select, for example, two AMVP prediction motion vectors for each of L0 and L1. Hereinafter, the AMVP prediction motion vector for L0 is referred to as “AMVP ₀₀ ” and “AMVP ₀₁ ”. The AMVP prediction motion vector for L1 is denoted as “AMVP ₁₀ ” and “AMVP ₁₁ ”. When there is no predicted motion vector that can be selected as the AMVP predicted motion vector among the predicted motion vectors of the blocks A0, A1, B0, B1, and B2, the compensation candidate generating unit 60 determines the predicted motion vector of the block C0. , May be selected as the temporal prediction motion vector. Further, when the predicted motion vector of the block C0 cannot be selected as the temporally predicted motion vector, the compensation candidate generating unit 60 may select the predicted motion vector of the block C1 as the temporally predicted motion vector. Hereinafter, the temporally-predicted motion vector for L0 is referred to as “temporally-predicted motion vector 0”. The temporal prediction motion vector for L1 is referred to as “temporal prediction motion vector 1”. The compensation candidate generation unit 60 may select one temporal prediction motion vector in addition to two AMVP prediction motion vectors for each of L0 and L1. The compensation candidate generating unit 60 may transmit the selected AMVP predicted motion vector or temporal predicted motion vector to the motion vector supplementing unit 10 for each of L0 and L1. The compensation candidate generating unit 60 may transmit the two selected AMVP predicted motion vectors and the temporal predicted motion vector to the motion vector supplementing unit 10 for each of L0 and L1. The compensation candidate generation unit 60 may further transmit a zero vector to the motion vector compensation unit 10.

  The mode determination unit 70 determines a motion vector compensation mode, which will be described later, and transmits a signal indicating the determined compensation mode to the motion vector compensation unit 10.

  Based on the value of the signal mDir and the compensation mode (for example, cMode in FIG. 3), the motion vector compensation unit 10 calculates the compensation vector cMV from the AMVP predicted motion vector or the temporal predicted motion vector received from the compensation candidate generation unit 60. Select The compensation mode cMode is input from, for example, the mode determination unit 70. The compensation mode cMode may be determined experimentally in advance, for example, and may be stored in a storage unit (not shown) outside the motion compensation unit 100. The motion vector compensation unit 10 may read the compensation mode cMode from the storage unit outside the motion compensation unit 100. The output of the motion vector compensation unit 10 is connected to one of the two input terminals of the first object selection unit 21 and one of the two input terminals of the second object selection unit 22 included in the object selection unit 20. .

One of the two input terminals of the first target selection unit 21 is connected to the motion vector compensation unit 10. The other two input terminals of the first target selection unit 21 is connected to a signal line motion vector MMV ₀ is input. Further, the control terminal of the first target selection unit 21 is connected to the signal mDir. The first target selection unit 21, depending on the value of the signal mdir, an input terminal connected to an output terminal, an input terminal connected to a signal line motion vector MMV ₀ is input, the motion vector interpolator 10 Switch between the signal line connected to. Specifically, the value of the signal mDir is, when a L0 or Bi described above, the first target selection unit 21 includes an input terminal connected to a signal line motion vector MMV ₀ is input, and an output terminal Connect between. When the value of the signal mDir is L1 described above, the first target selection unit 21 connects between the input terminal connected to the motion vector compensation unit 10 and the output terminal. The output terminal of the first object selection unit 21 is connected to the first refine unit 31 of the refine unit 30.

One of the two input terminals of the second target selection unit 22 is connected to the motion vector compensation unit 10. The other two input terminals of the second target selection unit 22 is connected to a signal line motion vector MMV ₁ is input. Further, the control terminal of the second target selection unit 22 is connected to the signal mDir. The second target selection unit 22, in accordance with the value of the signal mdir, an input terminal connected to an output terminal, an input terminal connected to a signal line motion vector MMV ₀ is input, the motion vector interpolator 10 Switch between the signal line connected to. Specifically, the value of the signal mDir is, if the above L1 or Bi, the first target selection unit 21 includes an input terminal connected to a signal line motion vector MMV ₁ is input, and an output terminal Connect between. When the value of the signal mDir is L0, the second target selection unit 22 connects between the input terminal connected to the motion vector compensation unit 10 and the output terminal. The output terminal of the second target selection unit 22 is connected to the second refinement unit 32 of the refinement unit 30.

The first refiner 31 generates a region (that is, a predicted image) in which the encoding cost of the encoding target block is minimum in a predetermined range based on the motion vector mMV ₀ of the reference image referenced by the input motion vector mMV ₀ . ) Is detected. The encoding cost may be any encoding cost selected from existing encoding costs. The encoding cost may be, for example, the sum of the sum of absolute values of errors between the encoding target block and the predicted image (SAD; Sum of Absolute Difference) and the weighted bit amount of the motion vector. Predetermined range based on the motion vector MMV _0, for example, may be an internal predetermined radius of a circle around the position indicated by the motion vector MMV _0. Predetermined range based on the motion vector MMV _0, for example, may be an internal graphic having a predetermined shape around the position indicated by the motion vector MMV _0. Predetermined range based on the motion vector MMV _0, for example, may be the entire reference image referred to by the motion vector MMV _0. The first refiner 31 sets a vector from the coordinates of the current block in the picture including the current block to the coordinates of a point indicating an area where the current cost is minimum as a predicted motion vector rMV _0. To detect. The first refiner 31 outputs, to the prediction selector 50, the detected image of the region where the encoding target cost is the minimum as a prediction signal “pred (rMV ₀ )” representing the prediction of the encoding target region. The first refining section 31, when the signal mDir is L1, i.e., if the inherited motion vector MMV ₀ not inputted, does not operate.

Second refining section 32, the reference image referred to by the motion vector MMV ₁ entered, in a predetermined range based on the motion vector MMV _1, the coding cost of the encoding target block is minimized area (i.e. the prediction image ) Is detected. Predetermined range based on the motion vector MMV ₁ may be, for example, inside the predetermined radius of a circle around the position indicated by the motion vector MMV _1. Predetermined range based on the motion vector MMV _1, for example, may be an internal graphic having a predetermined shape around the position indicated by the motion vector MMV _1. Predetermined range based on the motion vector MMV ₁ may be, for example, the whole reference image motion vector MMV ₁ refers. The second refiner 32 calculates a vector from the coordinates of the coding target block in the picture including the coding target block to the coordinates of a point indicating an area where the coding target cost is minimized, as a predicted motion vector rMV _1. To detect. The second refiner 32 outputs, to the prediction selector 50, the detected image of the region where the encoding target cost is minimized, as a prediction signal “pred (rMV ₁ )” representing the prediction of the encoding target region. Note that the second refining section 32, when the signal mDir is L2, i.e., if the inherited motion vector MMV ₁ is not inputted, does not operate.

When the value of the signal mDir is Bi, that is, when the inherited motion vectors mMV ₀ and mMV ₁ are input, the bi-prediction generation unit 40 generates bi-prediction based on the inherited motion vectors mMV ₀ and mMV ₁ . The bi-prediction generation unit 40 outputs the generated bi-prediction signal bipred (rMV ₀ , rMV ₁ ) representing the bi-prediction to the prediction selection unit 50. In the following description, the reference picture referred to by inherited motion vector MMV ₀ input, inherited motion vector MMV ₀ area indicated points selected from a predetermined range based on, referred to as a first reference area. Reference picture referenced by the input inherited motion vector MMV _1, inherited motion vector MMV ₁ area indicated points selected from a predetermined range based on, referred to as a second reference block. The bi-prediction generation unit 40 calculates the average image having the minimum encoding cost among the average images of the first reference region and the second reference region having the minimum encoding cost of the encoding target block as described above. Detect as bi-prediction. Then, the bi-prediction generation unit 40 specifies the first reference region and the second reference region in which the average image with the minimum coding cost is calculated. Further, bi-prediction generation section 40, the coordinates representing the current block, the motion vector to the coordinates of a first reference area specified, the predicted motion vector RMV _0. Similarly, bi-prediction generator 40, the coordinates representing the current block, the motion vector to the coordinates of a second reference area specified, the predicted motion vector RMV _1. Bi-prediction generator 40, if the inherited motion vector MMV ₀ or inherited motion vector MMV ₁ is not inputted, it does not operate.

  Note that one of the inherited motion vectors mMV0 and mMV1 is a picture temporally earlier than the encoding target picture, and the other reference picture is a temporally preceding picture than the encoding target picture. Is also good. In this case, among the average images in the first reference region and the second reference region in which the encoding cost of the encoding target block is the minimum, the average image with the minimum encoding cost is called bidirectional prediction. . In other words, the bi-prediction generation unit 40 may derive bi-prediction instead of bi-prediction, and output the derived bi-prediction.

The prediction selecting unit 50 outputs a code among the predicted image represented by the predicted signal pred (rMV ₀ ), the predicted image represented by the predicted signal pred (rMV ₁ ), and the predicted image represented by the bi-prediction signal bipred (rMV ₀ , rMV ₁ ). The prediction image that minimizes the cost to be converted is specified. Then, a prediction signal pred representing the specified prediction image and a prediction motion vector representing the prediction image represented by the prediction signal pred are output. The prediction selecting unit 50 further outputs the signal mDir as the signal rDir. In the following description, the signal rDir may also be described as “prediction direction”. When the signal representing the selected predicted image is the predicted signal pred (rMV ₀ ), the prediction selecting unit 50 outputs rMV ₀ as the predicted motion vector. When the signal representing the selected predicted image is the predicted signal pred (rMV ₁ ), the prediction selecting unit 50 outputs rMV ₁ as the predicted motion vector. When the signal representing the selected prediction image is the bi-prediction signal bipred (rMV ₀ , rMV ₁ ), the prediction selection unit 50 outputs rMV ₀ and rMV ₁ as prediction motion vectors.

With the above configuration of the motion compensating unit 100, when the inherited motion vector mMV ₀ is not input to the first refiner 31, the supplementary motion vector cMV is input to the first refiner 31. Further, when the inherited motion vector mMV ₁ is not input to the second refiner 32, the supplementary motion vector cMV is input to the second refiner 32.

In this comparative example, two vectors are input to the first refiner 31 and the second refiner 32. The first refinement unit 31 and the second refinement unit 32 operate without being in the pause state regardless of whether the picture including the current block is a B picture or a P picture. Further, in this comparative example, when the picture including the current block is a B picture, the bi-prediction generation unit 40 operates without being in the pause state. In this comparative example, the bi-prediction generation unit 40 enters a pause state when the picture including the current block is a P-picture. Next, the motion vector compensation unit 10 of this comparative example will be described with reference to the drawings. This will be described in detail.

  FIG. 6 is a block diagram schematically illustrating an example of the configuration of the motion vector compensation unit 10 of the comparative example. Referring to FIG. 6, the motion vector compensation unit 10 includes a detection unit 11, a compensation vector selection unit 12, a transmission unit 13, and a reception unit 14.

  The detection unit 11 detects whether the number of inherited motion vectors input to the refine unit 30 is two or one. For example, the detection unit 11 may receive the signal mDir from the pre-analysis unit 102. Then, when the value of the signal mDir is L0 or L1, the detection unit 11 may detect that the number of inherited motion vectors input to the refinement unit 30 is one. When the value of the signal mDir is Bi, the detection unit 11 only needs to detect that the number of inherited motion vectors input to the refinement unit 30 is two. The detection unit 11 may receive PicType, which is a signal indicating the type of the picture.

Receiving unit 14, the pre-analysis unit 102 receives at least one of the Inheritance motion vectors MMV ₀ and inheritance motion vector MMV _1. The receiving unit 14 further receives the compensation mode cMode from the mode determination unit 70. The receiving unit 14 further receives, for example, two selected predicted motion vectors (AMVP ₀₀ , AMVP ₀₁ , AMVP ₁₀ and AMVP ₁₁ ) and a temporally predicted motion vector (temporally predicted motion vector 0 and AMVP ₁₁ ) from the compensation candidate generating unit 60. And a temporal prediction motion vector 0).

When, for example, one input succession motion vector is input, the compensation vector selection unit 12 selects a compensation motion vector from the prediction motion vector and the temporal prediction motion vector according to the selection method in the compensation mode cMode. At that time, compensation vector selection unit 12, input inherited motion vector (i.e., inherited motion vector MMV ₀ or inherited motion vector MMV ₁₎ on the basis of, filling a motion vector may be selected. An example of the compensation mode and a selection method in the compensation mode will be described later in detail.

  Hereinafter, the motion vector compensation unit 10 will be described in more detail.

  FIG. 4 is a block diagram schematically illustrating an example of the configuration of the motion vector compensation unit 10 of the comparative example. FIG. 4 shows a configuration based on an example of implementation of the motion vector compensation unit 10 by a circuit. Referring to FIG. 4, the motion vector compensation unit 10 includes a first candidate selection unit 15, a second candidate selection unit 16, and a selection execution unit 17.

As described above, the compensation candidate generating unit 60 transmits at least one of the AMVP predicted motion vector and the temporal predicted motion vector and the zero vector to the motion vector supplementing unit 10. As described above, AMVP predicted motion vector _is AMVP _00, AMVP _01, AMVP ₁₀ and AMVP _11. The temporal prediction motion vector is a temporal prediction motion vector 0 and a temporal prediction motion vector 1. In the following description, a set of motion vectors transmitted from the compensation candidate generation unit 60 to the motion vector compensation unit 10 is referred to as “set A”. The set A includes, for example, AMVP ₀₀ , AMVP ₀₁ , AMVP ₁₀ , AMVP ₁₁ , temporally predicted motion vector 0, temporally predicted motion vector 1, and zero vector.

Among the vectors included in the set A, the AMVP predicted motion vector (AMVP ₀₀ and AMVP ₀₁ ) and the temporal predicted motion vector (temporally predicted motion vector 0) for L0 and the zero vector are input to the first candidate selecting unit 15. Is done. A set of vectors input to the first candidate selection unit 15 is referred to as “set 0”. Set 0 includes, for example, AMVP ₀₀ , AMVP ₀₁ , temporal prediction motion vector 0, and zero vector.

Among the vectors included in the set A, the AMVP predicted motion vector (AMVP ₁₀ and AMVP ₁₁ ) and the temporal predicted motion vector (temporally predicted motion vector 1) for L1 and the zero vector are input to the second candidate selection unit 16. Is done. The set of vectors input to the second candidate selection unit 16 is referred to as “set 1”. Set 1 includes, for example, AMVP ₁₀ , AMVP ₁₁ , temporal prediction motion vector 1 and zero vector.

  In addition, the inherited motion vector mMV0 received by the motion vector compensation unit 10 is input to the first candidate selection unit 15. The inherited motion vector mMV1 received by the motion vector compensation unit 10 is input to the second candidate selection unit 16. The compensation mode cMode received by the motion vector compensation unit 10 is input to the first candidate selection unit 15, the second candidate selection unit 16, and the selection execution unit 17.

  The portion (including the interface, for example) of the motion vector compensation unit 10 that receives the set A, the inherited motion vectors mMV0 and mMV1, and the compensation mode cMode input to the motion vector compensation unit 10 (including an interface, for example) is the reception unit 14 described above. Is equivalent to The part that transmits the set 0, which is at least a part of the set A, the inherited motion vector mMV0, and the compensation mode cMode to the first candidate selection unit 15 also corresponds to the above-described reception unit 14. The part that transmits the set 1, which is at least a part of the set A, the inherited motion vector mMV1, and the compensation mode cMode to the second candidate selection unit 16 corresponds to the reception unit 14 described above. The portion that transmits the compensation mode cMode to the selection execution unit 17 also corresponds to the above-described reception unit 14.

  In other words, the receiving unit 14 receives the set A, the inherited motion vector, and the compensation mode. Then, the receiving unit 14 inputs the vector included in the set 0 among the vectors included in the set A, the inherited motion vector mMV0, and the compensation mode cMode to the first candidate selecting unit 15. The receiving unit 14 further inputs the vector included in the set 1 among the vectors included in the set A, the inherited motion vector mMV1, and the compensation mode cMode to the second candidate selecting unit 16. Further, the reception unit 14 transmits the compensation mode cMode to the selection execution unit 17.

  The mDir and PicType (that is, the signal indicating mDir and the signal indicating PicType) received by the motion vector compensation unit 10 are input to the selection execution unit 17. The selection execution unit 17 detects the value of mDir and the value of PicType (hereinafter, also referred to as the value of input mDir and the value of PicType) indicated by the input signals.

  The part of the motion vector compensation unit 10 that receives the signal indicating mDir and the signal indicating PicType input to the motion vector compensation unit 10 and inputs those signals to the selection execution unit 17 corresponds to the detection unit 11 described above. I do. Further, the part of the selection execution unit 17 that detects the value of mDir and the value of PicType indicated by those signals input to the selection execution unit 17 also corresponds to the detection unit 11 described above.

  The first candidate selection unit 15 receives the set 0, mMV0, and cMode. Then, the first candidate selecting unit 15 selects a supplementary motion vector cMV0 from the vectors included in the received set 0 based on the received cMode or the mMV0 and the cMode. The first candidate selection unit 15 transmits the selected cMV0 to the selection execution unit 17.

  The second candidate selection unit 16 receives the set 1, the mMV1, and the cMode. Then, the second candidate selection unit 16 selects a supplementary motion vector cMV1 from the vectors included in the received set 1 based on the received cMode or the mMV1 and the cMode. The second candidate selection unit 16 transmits the selected cMV1 to the selection execution unit 17.

  The selection execution unit 17 selects the supplementary motion vector cMV0 or the supplementary motion vector cMV1 as the supplementary motion vector cMV based on the value of mDir, the value of PicType, and the cMode. The selection execution unit 17 outputs the selected supplementary motion vector cMV. In other words, the selection execution unit 17 transmits the supplementary motion vector cMV to the target selection unit 20. As described above, the motion vector compensation unit 10 outputs the selected compensation motion vector cMV.

  The portion (for example, including the interface) of the motion vector compensation unit 10 that outputs the compensation motion vector cMV corresponds to the transmission unit 13 described above. In other words, the transmission unit 13 outputs the supplementary motion vector cMV. In other words, the transmission unit 13 transmits the supplementary motion vector cMV to the target selection unit 20.

  The first candidate selection unit 15, the second candidate selection unit 16, and the selection execution unit 17 (excluding the above-described detection unit 11) correspond to the compensation vector selection unit 12. The method of selecting the supplementary motion vectors cMV0, cMV1, and cMV by the first candidate selection unit 15, the second candidate selection unit 16, and the selection execution unit 17 will be described in detail below. Different values need only be assigned in advance as the values of “replenishment mode 0” to “replenishment mode 7” in the following description.

(1) Compensation mode 0
The compensation vector selection unit 12 does not need to generate the compensation vector. In this comparative example, the compensation vector selection unit 12 does not generate a compensation vector when the compensation mode cMode is “compensation mode 0”.

  The details will be described below. When the compensation mode cMode is “compensation mode 0”, the first candidate selecting unit 15 does not need to select the compensation vector cMV0. The first candidate selection unit 15 does not need to transmit the compensation vector cMV0 to the selection execution unit 17. The second candidate selection unit 16 need not select the compensation vector cMV1. The second candidate selection unit 16 does not need to transmit the compensation vector cMV1 to the selection execution unit 17. The selection execution unit 17 does not need to select the cMV. The selection execution unit 17 does not need to output the cMV.

(2) Compensation mode 1
The compensation vector selection unit 12 may select an AMVP prediction motion vector having the minimum distance from the inherited motion vector as the compensation vector. In the present comparative example, when the compensation mode cMode is “compensation mode 1”, the compensation vector selection unit 12 selects an AMVP predicted motion vector having the shortest distance from the inherited motion vector as the compensation vector. For example, when mDir is L0, the compensation vector selecting unit 12 selects the AMVP prediction motion vector having the smaller distance from the inherited motion vector mMV0 among the two AMVP prediction motion vectors for L0 (that is, AMVP00 and AMVP01). All you have to do is select a vector. When the number of AMVP prediction motion vectors for L0 is one, the compensation vector selection unit 12 may select the AMVP prediction motion vector. For example, when mDir is L1, the compensation vector selection unit 12 selects the AMVP prediction motion vector having the smaller distance from the inherited motion vector mMV1 among the two AMVP prediction motion vectors for L1 (that is, AMVP10 and AMVP11). All you have to do is select a vector. When the number of AMVP prediction motion vectors for L1 is one, the compensation vector selection unit 12 may select the AMVP prediction motion vector.

  The details will be described below. If the compensation mode cMode is “compensation mode 1”, the first candidate selecting unit 15 selects, from AMVP00 and AMVP01, an AMVP predicted motion vector having a small distance from the input inherited motion vector mMV0 as the compensation vector cMV0. Good. When the number of AMVP prediction motion vectors for L0 is one, the first candidate selecting unit 15 may select the AMVP prediction motion vector as the compensation vector cMV0. The first candidate selection unit 15 transmits the selected compensation vector cMV0 to the selection execution unit 17. When the inherited motion vector mMV0 does not exist (that is, when the inherited motion vector mMV0 is not input to the first candidate selecting unit 15), the first candidate selecting unit 15 does not need to select cMV0. In this case, the first candidate selection unit 15 need not transmit the cMV0.

  When the compensation mode cMode is “compensation mode 1”, the second candidate selecting unit 16 may select, from the AMVPs 10 and 11, an AMVP predicted motion vector having a small distance from the inputted compensation vector mMV1 as the compensation vector cMV1. . When the number of AMVP prediction motion vectors for L1 is one, the second candidate selecting unit 16 may select the AMVP prediction motion vector as the compensation vector cMV1. The second candidate selection unit 16 transmits the selected compensation vector cMV1 to the selection execution unit 17. When the inherited motion vector mMV1 does not exist (that is, when the inherited motion vector mMV1 is not input to the second candidate selecting unit 16), the second candidate selecting unit 16 does not need to select the supplementary vector cMV1. In this case, the second candidate selecting unit 16 does not need to transmit the compensation vector cMV1.

  The selection executing unit 17 may select the compensation vector cMV0 or the compensation vector cMV1 as the compensation vector cMV based on the value of the prediction direction mDir.

  For example, when the prediction direction mDir is L0, the inherited motion vector mMV0 is input, but the inherited motion vector mMV1 is not input. In this case, the inherited motion vector mMV1 is not input to the second candidate selection unit 16. Therefore, the second candidate selection unit 16 does not output the compensation vector cMV1. Further, the inherited motion vector mMV0 is input to the first candidate selection unit 15. The first candidate selecting unit 15 may select cMV0 based on the inherited motion vector mMV0 as described above, and output the selected cMV0. When the prediction direction mDir is L0, the selection execution unit 17 may select the compensation vector cMV0 as the compensation vector cMV.

  For example, when the prediction direction mDir is L1, the inherited motion vector mMV1 is input, but the inherited motion vector mMV0 is not input. In this case, the inherited motion vector mMV0 is not input to the first candidate selection unit 15. Therefore, the first candidate selection unit 15 does not output the compensation vector cMV0. Further, the inherited motion vector mMV1 is input to the second candidate selection unit 16. Then, the second candidate selecting unit 16 may select the compensation vector cMV1 based on the inherited motion vector mMV1 as described above, and output the selected compensation vector cMV1. When the prediction direction mDir is L1, the selection execution unit 17 may select the compensation vector cMV1 as the compensation vector cMV.

  When the prediction direction mDir is Bi, the number of inherited motion vectors is two. In this case, the selection execution unit 17 does not need to select the compensation vector cMV. The selection execution unit 17 does not need to output the compensation vector cMV.

(3) Compensation mode 2
The compensation vector selection unit 12 may select the AMVP prediction motion vector having the largest distance from the inherited motion vector as the compensation vector. In this comparative example, when the compensation mode cMode is “compensation mode 2”, the compensation vector selection unit 12 selects the AMVP predicted motion vector having the largest distance from the inherited motion vector as the compensation vector. For example, when mDir is L0, the compensation vector selecting unit 12 selects the AMVP prediction motion vector having a longer distance from the inherited motion vector mMV0 from the two AMVP prediction motion vectors for L0 (that is, AMVP00 and AMVP01). All you have to do is select a vector. When the number of AMVP prediction motion vectors for L0 is one, the compensation vector selection unit 12 may select the AMVP prediction motion vector. For example, when mDir is L1, the compensation vector selection unit 12 selects the AMVP prediction motion vector having the larger distance from the inherited motion vector mMV1 from the two AMVP prediction motion vectors for L1 (that is, AMVP10 and AMVP11). All you have to do is select a vector. When the number of AMVP prediction motion vectors for L1 is one, the compensation vector selection unit 12 may select the AMVP prediction motion vector.

  The details will be described below. When the compensation mode cMode is “compensation mode 1”, the first candidate selecting unit 15 may select an AMVP prediction motion vector having a large distance from the inputted compensation vector mMV0 as the compensation vector cMV0 from AMVP00 and AMVP01. . When the number of AMVP prediction motion vectors for L0 is one, the first candidate selecting unit 15 may select the AMVP prediction motion vector as the compensation vector cMV0. The first candidate selection unit 15 transmits the selected compensation vector cMV0 to the selection execution unit 17. When the compensation vector mMV0 does not exist (that is, when the compensation vector mMV0 is not input to the first candidate selection unit 15), the first candidate selection unit 15 does not need to select cMV0. In this case, the first candidate selection unit 15 need not transmit the cMV0.

  When the compensation mode cMode is “compensation mode 1”, the second candidate selecting unit 16 may select an AMVP prediction motion vector having a large distance from the inputted compensation vector mMV1 as the compensation vector cMV1 from AMVP10 and AMVP11. . When the number of AMVP prediction motion vectors for L1 is one, the second candidate selecting unit 16 may select the AMVP prediction motion vector as the compensation vector cMV1. The second candidate selection unit 16 transmits the selected compensation vector cMV1 to the selection execution unit 17. When the compensation vector mMV1 does not exist (that is, when the compensation vector mMV1 is not input to the second candidate selection unit 16), the second candidate selection unit 16 does not have to select the cMV1. In this case, the second candidate selection unit 16 need not transmit the cMV1.

  The selection execution unit 17 may select cMV0 or cMV1 as cMV according to the value of mDir. For example, when mDir is L0, mMV0 is input to the first candidate selection unit 15. The first candidate selection unit 15 selects the cMV0 based on the input mMV0, and transmits the selected cMV0 to the selection execution unit 17. When mDir is L0, mMV1 is not input to the second candidate selection unit 16. The second candidate selection unit 16 does not output cMV1. Therefore, when mDir is L0, the selection execution unit 17 may select cMV0 as cMV. For example, when mDir is L1, mMV1 is input to the second candidate selection unit 16. The second candidate selection unit 16 selects the cMV1 based on the input mMV1, and transmits the selected cMV1 to the selection execution unit 17. When mDir is L1, mMV0 is not input to the first candidate selection unit 15. Therefore, the first candidate selection unit 15 does not output cMV0. Therefore, when mDir is L1, the selection execution unit 17 may select cMV1 as cMV. If mDir is Bi, the number of inherited motion vectors is two. In this case, the selection execution unit 17 does not need to select the cMV. The selection execution unit 17 does not need to output the cMV.

(4) Compensation mode 3
The compensation vector selection unit 12 may select a zero vector as the compensation vector. In this comparative example, when the compensation mode cMode is “compensation mode 3”, the compensation vector selection unit 12 selects the zero vector as the compensation vector.

  The details will be described below. When the compensation mode cMode is “compensation mode 3”, one of the first candidate selection unit 15 and the second candidate selection unit 16 (for example, the first candidate selection unit 15) may select the zero vector as the compensation vector. . Then, the first candidate selecting unit 15 may transmit the selected zero vector to the selection executing unit 17 as the compensation vector cMV0. In this case, the other of the first candidate selecting unit 15 and the second candidate selecting unit 16 (for example, the second candidate selecting unit 16) does not need to operate and does not need to output the compensation vector. In this case, the selection execution unit 17 may transmit the zero vector transmitted from one of the first candidate selection unit 15 and the second candidate selection unit 16 as the compensation vector to the target selection unit 20 as the compensation vector cMV.

  Of the first candidate selection unit 15 and the second candidate selection unit 16, the one that selects the zero vector as the compensation vector when the compensation mode cMode is “complementation mode 3” is referred to as “operating candidate selection unit”. . The operating candidate selection unit may be fixed. The operating candidate selection unit may change according to the inherited motion vector input to the motion vector compensation unit 10. For example, when the compensation mode cMode is “compensation mode 3” and the inherited motion vector mMV0 is received, the first candidate selection unit 15 may select the zero vector as the compensation vector cMV0. When the compensation mode cMode is “compensation mode 3” and the prediction direction mDir is L0, the selection execution unit 17 may select cMV0 as cMV. For example, when the compensation mode cMode is “compensation mode 3” and the inherited motion vector mMV1 is received, the second candidate selection unit 16 may select a zero vector as the compensation vector cMV1. When the compensation mode cMode is “compensation mode 3” and the prediction direction mDir is L1, the selection execution unit 17 may select cMV1 as cMV.

  The above-described compensation modes from the compensation mode 0 to the compensation mode 3 can be selected whether the picture including the encoding target block is a B picture or a P picture.

  The compensation modes from compensation mode 4 to compensation mode 7 described below can be selected when the picture including the current block is a B picture. When the input compensation mode cMode is the compensation mode 4 to the compensation mode 7 and the input PicType is a P picture, the motion vector compensation unit 10 performs the compensation in any one of the compensation modes 0 to 3. May work. The compensation mode in that case may be determined in advance. In this case, for example, the selection execution unit 17 may transmit a predetermined compensation mode to the first candidate selection unit 15 and the second candidate selection unit 16. The flow of data (for example, signals) in this case is not shown. In this case, the first candidate selection unit 15 and the second candidate selection unit 16 may operate in the compensation mode received from the selection execution unit 17 instead of the compensation mode indicated by the inputted compensation mode cMode. In the compensation modes from the compensation mode 4 to the compensation mode 7, the compensation vector is compensated as a motion vector in a prediction direction different from the inherited motion vector.

(5) Compensation mode 4
The compensation vector selection unit 12 may select an AMVP prediction motion vector whose index is 0 as the compensation vector. In this comparative example, when the compensation mode cMode is “compensation mode 4”, the compensation vector selection unit 12 may select the AMVP prediction motion vector whose index is 0 as the compensation vector.

  The compensation candidate generating unit 60 of this comparative example may assign 0 as an index to one of the two AMVP predicted motion vectors and assign 1 to the other as an index for each prediction direction. Specifically, the compensation candidate generating unit 60 may, for example, assign 0 as an index to AMVP00 and AMVP10, and assign 1 as an index to AMVP01 and AMVP11. The compensation candidate generation unit 60 may transmit the AMVP predicted motion vector to which the index has been added to the motion vector compensation unit 10.

  In this comparative example, when the compensation mode is “compensation mode 4”, the compensation vector selection unit 12 selects, for example, a prediction having no inherited motion vector among two AMVP predicted motion vectors having an index of 0 as the compensation vector. What is necessary is just to select the AMVP prediction motion vector of a direction. For example, when mDir is L0, the compensation vector selecting unit 12 may select an AMVP prediction motion vector for L1 (that is, AMVP10) for which the assigned index is 0. For example, when mDir is L1, the compensation vector selection unit 12 may select an AMVP prediction motion vector for L0 (that is, AMVP00) for which the assigned index is 0.

  The details will be described below. When the compensation mode cMode is “compensation mode 4”, the first candidate selecting unit 15 selects the AMVP prediction motion vector for L0 to which the assigned index is 0 (that is, AMVP00) as cMV0. Then, the first candidate selection unit 15 transmits cMV0 to the selection execution unit 17. The second candidate selection unit 16 selects, as the cMV1, the AMVP prediction motion vector for L1 to which the assigned index is 0 (that is, AMVP10). Then, the second candidate selection unit 16 transmits the cMV1 to the selection execution unit 17.

  The selection execution unit 17 may select cMV0 or cMV1 as cMV based on the value of mDir. For example, when the value of mDir indicating the prediction direction of the inherited motion vector is L0, the selection execution unit 17 may select a compensation vector in a prediction direction different from L0, which is the value of mDir, as the compensation vector cMV. That is, the selection execution unit 17 may select the compensation vector cMV1 and the compensation vector cMV whose prediction directions are L1. For example, when the value of mDir indicating the prediction direction of the inherited motion vector is L1, the selection execution unit 17 may select a compensation vector in a prediction direction different from L1 that is the value of mDir as the compensation vector cMV. That is, the selection execution unit 17 may select the compensation vector cMV0 and the compensation vector cMV whose prediction directions are L0. The selection execution unit 17 may transmit the selected cMV to the target selection unit 20.

(6) Compensation mode 5
The compensation vector selection unit 12 may select an AMVP prediction motion vector whose index is 1 as the compensation vector. In this comparative example, when the compensation mode cMode is “compensation mode 5”, the compensation vector selection unit 12 selects the AMVP prediction motion vector whose index is 1 as the compensation vector. The compensation vector selection unit 12 may select, for example, an AMVP prediction motion vector in the prediction direction having no inherited motion vector from the two AMVP prediction motion vectors having the index 1 as the compensation vector. For example, when mDir is L0, the compensation vector selection unit 12 may select the AMVP prediction motion vector for L1 (that is, AMVP11), to which the assigned index is 1. For example, when mDir is L1, the compensation vector selection unit 12 may select the AMVP prediction motion vector for L0 (that is, AMVP01) to which the assigned index is 1.

  The details will be described below. When the compensation mode cMode is “compensation mode 5”, the first candidate selecting unit 15 selects the AMVP prediction motion vector for L0 to which the assigned index is 1 (that is, AMVP01) as the cMV0. Then, the first candidate selection unit 15 transmits cMV0 to the selection execution unit 17. In this case, the second candidate selection unit 16 selects, as the cMV1, the AMVP prediction motion vector for L1 to which the assigned index is 1 (that is, AMVP11). Then, the second candidate selection unit 16 transmits the cMV1 to the selection execution unit 17.

  The selection execution unit 17 may select cMV0 or cMV1 as cMV based on the value of mDir. For example, when the value of mDir indicating the prediction direction of the inherited motion vector is L0, the selection execution unit 17 may select a compensation vector in a prediction direction different from L0, which is the value of mDir, as the compensation vector cMV. That is, the selection execution unit 17 may select the compensation vector cMV1 and the compensation vector cMV whose prediction directions are L1. For example, when the value of mDir indicating the prediction direction of the inherited motion vector is L1, the selection execution unit 17 may select a compensation vector in a prediction direction different from L1 that is the value of mDir as the compensation vector cMV. That is, the selection execution unit 17 may select the compensation vector cMV0 and the compensation vector cMV whose prediction directions are L0. The selection execution unit 17 may transmit the selected cMV to the target selection unit 20.

(7) Compensation mode 6
The compensation vector selection unit 12 may select a zero vector as the compensation vector. In this comparative example, when the compensation mode cMode is “compensation mode 6”, the compensation vector selection unit 12 selects the zero vector as the compensation vector.

  The details will be described below. When the compensation mode cMode is “compensation mode 6”, the first candidate selecting unit 15 selects the zero vector as cMV0. The first candidate selection unit 15 transmits the zero vector to the selection execution unit 17 as cMV0. When the compensation mode cMode is “compensation mode 6”, the second candidate selection unit 16 selects the zero vector as cMV1. The second candidate selection unit 16 transmits the zero vector to the selection execution unit 17 as cMV1.

  When the compensation mode cMode is the “compensation mode 6”, the selection execution unit 17 outputs either cMV0 or cMV1 as cMV. The selection execution unit 17 may be configured to output cMV0 as cMV when the compensation mode cMode is “compensation mode 6”. The selection execution unit 17 may be configured to output cMV1 as cMV when the compensation mode cMode is the “compensation mode 6”. The selection execution unit 17 may select cMV0 or cMV1 according to the value of mDir. For example, the selection executing unit 17 may select cMV1 when the value of mDir is L0, and may select cMV0 when the value of mDir is L1. The relationship between the value of mDir and cMV0 and cMV1 may be reversed. Since cMV0 and cMV1 are zero vectors, cMV is also a zero vector.

(8) Compensation mode 7
The compensation vector selection unit 12 may select a temporal prediction motion vector as the compensation vector. In this comparative example, when the compensation mode cMode is “compensation mode 7”, the compensation vector selection unit 12 selects a temporal prediction motion vector as the compensation vector.

  As described above, the compensation candidate generation unit 60 selects a temporal prediction motion vector for each prediction direction. That is, the compensation candidate generation unit 60 compares the temporal prediction motion vector 0 that is the temporal prediction motion vector for the prediction direction L0 and the temporal prediction motion vector 1 that is the temporal prediction motion vector for the prediction direction L1 as follows, for example. To choose. For example, when the predicted motion vector in the block C0 in the example illustrated in FIG. 5 exists, the compensation candidate generation unit 60 selects the predicted motion vector in the block C0 as the temporal predicted motion vector. If there is no predicted motion vector in the block C0, the compensation candidate generating unit 60 selects the predicted motion vector in the block C1 as the temporal predicted motion vector. When there is no predicted motion vector in the block of C1, the compensation candidate generating unit 60 sets the temporally predicted motion vector to a zero vector.

  Specifically, when the compensation mode cMode is “compensation mode 7” and the prediction direction mDir is L0, the compensation vector selection unit 12 sets the temporal prediction vector for the prediction direction L1 (that is, the temporal prediction vector 1). Select as the compensation vector cMV. When the compensation mode cMode is “compensation mode 7” and the prediction direction mDir is L1, the compensation vector selection unit 12 selects a temporal prediction vector for the prediction direction L0 (that is, the temporal prediction vector 0) as the compensation vector cMV. I do.

  The details will be described below. When the compensation mode cMode is “compensation mode 7”, the first candidate selecting unit 15 selects a temporal prediction vector in the prediction direction L0 (that is, the temporal prediction vector 0) as cMV0. The first candidate selection unit 15 transmits the selected cMV0 to the selection execution unit 17. When the compensation mode cMode is “compensation mode 7”, the second candidate selection unit 16 selects the temporal prediction vector in the prediction direction L1 (that is, the temporal prediction vector 1) as cMV1. The first candidate selection unit 15 transmits the selected cMV1 to the selection execution unit 17.

  When the compensation mode cMode is “compensation mode 7” and the prediction direction mDir is L0, the selection execution unit 17 selects cMV1 as cMV. When the compensation mode cMode is “compensation mode 7” and the prediction direction mDir is L1, the selection execution unit 17 selects cMV0 as cMV.

  Next, the operation of this comparative example will be described in detail with reference to the drawings.

  FIG. 8 is a flowchart illustrating an example of an operation of generating a predicted image of an encoding target image by the video encoding device 1 of the comparative example. When the operation shown in FIG. 8 is started, the inherited motion vector has been generated by the pre-analysis unit 102. The pre-analysis unit 102 transmits the inherited motion vector (that is, at least one of the inherited motion vectors mDir0 and mDir1), the prediction direction mDir, and the picture type PicType to the motion vector compensation unit 10.

  Referring to FIG. 8, first, the mode determination unit 70 determines a compensation mode (step S101). When the compensation mode is fixed, the mode determination unit 70 does not need to determine the compensation mode. The mode determination unit 70 transmits the compensation mode to the motion vector compensation unit 10. Next, the compensation candidate generation unit 60 generates an AMVP prediction motion vector and a temporal prediction motion vector (Step S102). The compensation candidate generating unit 60 transmits the set A described above, which is a set of vectors including the generated AMVP predicted motion vector and the temporal predicted motion vector, to the motion vector supplementing unit 10. Next, the motion vector compensation unit 10 executes a motion vector compensation process (step S103). The motion vector compensation processing will be described later in detail. When the number of inherited motion vectors is one, the motion vector compensation unit 10 transmits the inherited motion vector and the compensated motion vector compensated by the motion vector compensation process to the refine unit 30. Next, the refiner 30 refines the two received motion vectors (step S104). That is, the refiner 30 detects a predicted motion vector based on the two received motion vectors. The two received motion vectors are, as described above, two inherited motion vectors, or an inherited motion vector and a supplementary motion vector. The refiner 30 further generates a predicted image based on the predicted motion vector obtained (that is, detected) by the refinement (step S105).

  When the picture including the current block is a P picture (NO in step S106), the video encoding device 1 next performs an operation of step S108. In this case, the prediction selecting unit 50 selects, for example, one predicted image having a smaller encoding cost from the two predicted images. If the picture including the current block is a B picture (YES in step S106), then the bi-prediction generation unit 40 generates a bi-prediction image (step S107). Then, the prediction selecting unit 50 selects a predicted image (Step S108). In this case, the prediction selecting unit 50 selects one image as a predicted image from two predicted images and one bi-predicted image. Then, the prediction selection unit 50 outputs a prediction vector indicating the selected prediction image, the selected prediction image, and a prediction direction indicated by the selected prediction image (Step S109).

  Next, the operation of the motion vector compensation processing of the video encoding device 1 of the present comparative example will be described in detail with reference to the drawings.

  FIG. 9 is a flowchart illustrating an example of an operation of a motion vector compensation process of the video encoding device 1 of the comparative example.

  Referring to FIG. 9, first, the detection unit 11 of the motion vector compensation unit 10 (for example, the selection execution unit 17 illustrated in FIG. 4 that operates as the detection unit 11) illustrated in FIG. 6 is input to the motion compensation unit 100. Then, the number of inherited motion vectors is detected (step S201). Specifically, the detection unit 11 detects the value of the signal mDir. Then, the detecting unit 11 detects whether the number of inherited motion vectors is one or two based on the value of the signal mDir. When the value of the signal mDir is L0 or L1, the detection unit 11 detects that the number of inherited motion vectors is one. When the value of the signal mDir is Bi, the detection unit 11 detects that the number of inherited motion vectors is two.

  If the number of inherited motion vectors is not one (NO in step S202), that is, if the value of the signal mDir is Bi, the video encoding device 1 ends the motion vector compensation processing shown in FIG.

  When the number of inherited motion vectors is one (YES in step S202), that is, when the value of the signal mDir is L0 or L1, the receiving unit 14 receives the compensation mode (step S203). Specifically, the motion vector compensation unit 10 receives the compensation mode cMode. The first candidate selection unit 15, the second candidate selection unit 16, and the selection execution unit 17 receive the compensation mode cMode received by the motion vector compensation unit 10. When the compensation mode is determined in advance, the receiving unit 14 does not need to receive the compensation mode.

  Next, the receiving unit 14 of the motion vector compensation unit 10 receives the vector included in the set A (that is, the vector of the set A) from the compensation candidate generation unit 60 (Step S204). Specifically, the vectors included in set A are AMVP00, AMVP01, AMVP10, AMVP11, temporally predicted motion vector 0, temporally predicted motion vector 1, and zero vector. Set A is a set of the above vectors. The motion vector compensation unit 10 may not receive a vector that is not generated by the compensation candidate generation unit 60 among the vectors included in the set A. The first candidate selection unit 15 receives AMVP00, AMVP01, a temporal prediction motion vector 0, and a zero vector (ie, set 0) among the vectors of the set A received by the motion vector compensation unit 10. The first candidate selection unit 15 may not receive a vector that is not generated by the compensation candidate generation unit 60 among the vectors of the set 0. The second candidate selection unit 16 receives the AMVP 10, the AMVP 11, the temporally predicted motion vector 1, and the zero vector (ie, set 1) among the vectors of the set A received by the motion vector compensation unit 10. The second candidate selection unit 16 may not receive a vector not generated by the compensation candidate generation unit 60 among the vectors of the set 1.

  The receiving unit 14 further receives the inherited motion vector (Step S205). Specifically, the motion vector compensation unit 10 receives the inherited motion vector mMV0 and the inherited motion vector mMV1. The first candidate selection unit 15 receives the inherited motion vector mMV0 among the inherited motion vectors received by the motion vector compensation unit 10. The second candidate selection unit 16 receives the inherited motion vector mMV1 among the inherited motion vectors received by the motion vector compensation unit 10. The motion vector compensation unit 10 does not have to receive a vector that is not generated by the pre-analysis unit 102 among the inherited motion vector mMV0 and the inherited motion vector mMV1. If the inherited motion vector mMV0 has not been generated by the pre-analysis unit 102, the first candidate selecting unit 15 need not receive the inherited motion vector mMV0. When the inherited motion vector mMV1 has not been generated by the pre-analysis unit 102, the second candidate selecting unit 16 does not need to receive the inherited motion vector mMV1.

  Next, the compensation vector selection unit 12 selects a compensation motion vector from at least one of the received AMVP predicted motion vector, temporal predicted motion vector, and zero vector based on the received compensation mode (step S206). ). Specifically, the first candidate selecting unit 15 selects the compensation vector cMV0 from the vectors included in the received set 0 based on the received cMode as described above. The second candidate selection unit 16 selects the compensation vector cMV1 from the vectors included in the received set 0 based on the received cMode as described above. As described above, the selection execution unit 17 selects the compensation vector cMV from cMV0 and cMV1, based on the prediction direction mDir and the compensation mode cMode.

  The transmission unit 13 transmits the selected supplementary motion vector cMV to the target selection unit 20 via the target selection unit 20 (Step S207). Specifically, the selection execution unit 17 transmits the selected supplementary vector cMV to the target selection unit 20. Then, the video encoding device 1 ends the operation illustrated in FIG.

  In the above operation example, when the number of inherited motion vectors is 2, the motion vector compensation unit 10 does not perform the processing from step S203 to step S207. However, regardless of the number of inherited motion vectors, the motion vector compensation unit 10 generates the compensation motion vector by performing the operations from step S203 to step S207, and transmits the generated compensation motion vector to the target selection unit 20. May be. In this case, when the number of inherited motion vectors is 2, the target selecting unit 20 may select those inherited motion vectors as the motion vectors before refinement. Then, in this case, the target selection unit 20 only needs to select the received supplementary motion vector as the motion vector before refinement.

  The comparative example described above has a first effect that the accuracy of prediction can be improved even when the number of reference pictures is one.

  The reason is that, when the number of reference pictures is one, that is, when the number of inherited motion vectors is one, the compensation motion vector selected by the motion vector compensation unit 10 is supplied to the refine unit 30. In this case, the prediction selecting unit 50 selects, for example, an image with the smallest encoding cost from the predicted image based on the inherited motion vector and the predicted image based on the supplementary motion vector. Alternatively, the prediction selecting unit 50 selects, for example, an image having the smallest encoding cost from the predicted image based on the inherited motion vector, the predicted image based on the supplementary motion vector, and the bi-predicted image. Therefore, the prediction accuracy of the selected image for the current block is improved.

  Further, it is expected that the accuracy of the motion compensation prediction can be improved by the additional evaluation of the motion vector, which cannot be performed by the pre-analysis unit 102 (generally called a pre-analyzer). The prediction directions and components of the inherited motion vectors mMV0 and mMV1 determined by the pre-analysis unit 102 are not always optimal. In particular, since the motion vector search in consideration of the temporal direct mode is not performed in the B picture, it is highly likely that the selection of the prediction direction is not optimal.

  Furthermore, overhead reduction can be expected by selecting motion vectors suitable for spatio-temporal adaptive prediction of motion vectors in the HEVC standard (that is, AMVP and merge prediction). The inherited motion vectors mMV0 and mMV1 determined by the pre-analyzer based on the median prediction are not necessarily motion vectors optimal for AMVP and merge prediction.

  This comparative example further has a second effect that the arithmetic unit (that is, the first refiner 31, the second refiner 32, and the bi-prediction generator 40 shown in FIG. 3) can be used. . The reason is that, when there is one inherited motion vector, the supplementary motion vector selected by the supplementary vector selecting unit 12 is the input of the inherited motion vector of the first refinement unit 31 and the second refinement unit 32. This is because it is input to those who do not. Furthermore, if the picture including the current block is a B picture, the bi-prediction generation unit 40 generates bi-prediction even when the number of inherited motion vectors is one.

  FIG. 7 is a diagram illustrating an example of a structure of a portion that performs motion vector compensation and refinement in the video encoding device 1 according to the comparative example. In FIG. 7, mcMV0 and mcMV1 represent two pre-refining motion vectors selected by the target selection unit 20 among mMV0, mMV1, and cMV.

The vector compensation described above is, in a simplified manner, performed by a device having the structure of FIG. Hereinafter, the structure illustrated in FIG. 7 is referred to as an “ideal structure”. In an ideal structure, vector compensation and motion search processing are collectively executed for each block. That is, when vector compensation is performed on a certain block, immediately after that, a motion search process is performed by the motion search process centering on the inherited motion vector and the supplementary motion vector. Among the motion vectors obtained by the motion search, a better motion index is selected. The selected motion vector becomes a new motion vector for the block and is used when generating a supplementary motion vector for the subsequent block. Here, the fact that the selected motion vector is used when generating a supplementary motion vector for a subsequent block is referred to as “vector propagation”.

  In general, for example, in order to satisfy a condition for a processing speed, a restriction may be imposed on the processing. As described above, for example, realizing a process that satisfies a condition by adding a constraint to the process may be referred to as “restricted realization”. For example, in the above-described motion vector compensation in the comparative example, one of the restrictions is to interrupt the continuity of the process between the process of supplementing the vector and the process of refinement. For example, the continuity of the processing can be cut off between the processing of supplementing the vector and the processing of the refinement as follows.

(1) Arranging a storage unit between the object selection unit 20 and the refine unit 30 (2) Storing the selected motion vector in the storage unit by the object selection unit 20 (3) Refining the stored motion vector by the refiner 30 By cutting off the continuity of the processing described above, the following can be performed in parallel for a plurality of blocks.

(1) Performing motion vector compensation (2) Performing refinement processing based on motion vectors supplemented for a plurality of blocks different from the block to be compensated for motion vectors In the comparative example described above. In order to generate an AMVP vector used for interpolation, a latest motion vector for a peripheral block (for example, a block at a position of A0 to C1 shown in FIG. 5) is required. Then, the process of compensating for the motion vector and the process of refinement can operate in parallel for a plurality of blocks. Therefore, there is a possibility that a process of compensating for a motion vector is started at a stage where the process of refining a certain block has not been completed.

  At this time, in the refinement process, it is not possible to use a motion vector generated by the refinement process executed in parallel with the refinement process (hereinafter, also referred to as “motion vector after refinement”). However, instead of the motion vector generated by the refinement process, a motion vector used in the refinement process for generating the motion vector (hereinafter, also referred to as “motion vector before refinement”) can be used. is there. When the above-described storage unit is implemented, the motion vector before refinement is stored in the storage unit. The refinement unit can read the motion vector before the refinement from the storage unit and use the readout motion vector before the refinement.

  The above-mentioned storage unit stores, as a motion vector before refinement, an inherited motion vector that has not been refined and a supplementary motion vector. There is no means for determining which of the inherited motion vector and the supplementary motion vector is suitable for use as the refined motion vector. However, there is a high possibility that the inherited motion vector obtained by the pre-analysis unit 102 is more suitable than the supplementary motion vector selected from the vectors of the peripheral blocks. Therefore, in the above-described “realization with restriction” of the comparative example, an inherited motion vector that has not been refined may be used as a motion vector after refinement of a peripheral block. A supplementary motion vector selected from a vector of a peripheral block of the current block need not be used as a candidate for a supplementary motion vector of the current block.

  As described above, in the above-described restricted realization of the comparative example, before the motion vectors of the peripheral blocks are determined by the refinement process, there is a possibility that the compensation of the motion vectors is started. If the motion vector compensation is started before the motion vector of the peripheral block is determined by the refinement process, the motion vector before refinement is used as a candidate of the motion vector to be compensated. Does not work.

  Next, embodiments of the present invention will be described in detail with reference to the drawings.

[First Embodiment]
[Description of configuration]
A first embodiment of the present invention will be described in detail with reference to the drawings.

  FIG. 10 is a block diagram illustrating an example of a configuration of the video encoding device 2 according to the present embodiment.

  Comparing the configuration of the video encoding device 2 shown in FIG. 10 with the configuration of the video encoding device 1 according to the comparative example shown in FIG. 1, the video encoding device 2 The difference is that a compensator 200 is included.

  FIG. 11 is a block diagram schematically illustrating an example of the configuration of the motion compensation unit 200 of the video encoding device 2 according to the present embodiment.

  Comparing the configuration of the motion compensation unit 200 shown in FIG. 11 with the configuration of the motion compensation unit 100 shown in FIG. 2, the motion compensation unit 200 includes the motion vector 90 instead of the motion vector compensation unit 10, and further stores the vector. Unit 80.

  The target selection unit 20 of the present embodiment generates a pre-refinement motion vector as in the above-described comparative example. The target selection unit 20 of the present embodiment stores the motion vector before refinement in the vector storage unit 80 instead of transmitting the motion vector before refinement to the refinement unit 30. That is, when the pre-analysis unit 102 generates two inherited motion vectors, the target selecting unit 20 stores the two inherited motion vectors in the vector storage unit 80 as the pre-refining motion vector. When the pre-analyzing unit 102 generates one inherited motion vector, the target selecting unit 20 uses the inherited motion vector and the supplementary motion vector generated by the motion vector compensating unit 90 as a motion vector before refinement, as a vector storage unit. 80.

  The vector storage unit 80 stores the motion vector before refinement generated by the target selection unit 20.

  The compensation candidate generation unit 60 according to the present embodiment generates an AMVP prediction motion vector and a temporal prediction motion vector based on the motion vectors before refinement of the neighboring blocks of the encoding target block stored in the vector storage unit 80. As described above, the peripheral blocks are, for example, blocks A0 to C1 illustrated in FIG. The compensation candidate generation unit 60 uses the vector stored in the vector storage unit 80 among the motion vectors before refinement of these blocks as the prediction motion vector of the peripheral block, and uses the AMVP prediction motion vector and the temporal prediction motion vector. Should be generated. The method of generating the AMVP prediction motion vector and the temporal prediction motion vector by the compensation candidate generation unit 60 is, for example, the same as the general method of generating the AMVP prediction motion vector and the temporal prediction motion vector based on the prediction motion vector of the peripheral block. Method is good.

  The motion vector compensation unit 90 generates a compensation motion vector based on the AMVP prediction motion vector, the temporal prediction motion vector, and the zero vector generated by the compensation candidate generation unit 60. When the number of inherited motion vectors generated by the pre-analysis unit 102 is one, the motion vector compensation unit 90 transmits the generated compensation motion vector to the target selection unit 20.

  The refinement unit 30 reads the pre-refinement motion vector from the vector storage unit 80, and refines the read-out pre-refinement motion vector in the same manner as the refinement unit 30 of the comparative example.

  The bi-prediction generation unit 40 of the present embodiment is the same as the bi-prediction generation unit 40 of the comparative example described above. The prediction selection unit 50 of the present embodiment is the same as the prediction selection unit 50 of the comparative example described above.

  Hereinafter, the motion compensation unit 200 of the present embodiment will be described in more detail.

  FIG. 12 is a block diagram illustrating an example of a configuration of the motion compensation unit 200 according to the present embodiment.

  When FIG. 12 is compared with FIG. 3 illustrating an example of the configuration of the motion compensation unit 100 according to the comparative example, the motion compensation unit 200 according to the present embodiment includes a motion vector compensation unit 90 instead of the motion vector compensation unit 10. In addition, a vector storage unit 80 is connected between the target selection unit 20 and the refine unit 30. The vector storage unit 80 includes a first vector storage unit 81 and a second vector storage unit 82.

  The first target selection unit 21 stores the selected pre-refinement motion vector in the first vector storage unit 81. The second target selection unit 22 stores the selected pre-refinement motion vector in the second vector storage unit 82. The first target selection unit 21 and the second target selection unit 22 are the same as the first target selection unit 21 and the second target selection unit 22 of the above-described comparative example, respectively, except for the above differences.

  The first refiner 31 reads a motion vector before refinement from the first vector storage unit 81 and refines the read motion vector before refinement to generate a predicted motion vector rMV0. The second refiner 32 reads a motion vector before refinement from the second vector storage unit 82 and refines the read motion vector before refinement to generate a predicted motion vector rMV1. The first refiner 31 and the second refiner 32 are the same as the first refiner 31 and the second refiner 32 of the above-described comparative example, except for the above difference. That is, the first refiner 31 of the present embodiment performs the refinement process in the same manner as the first refiner 31 of the comparative example, and outputs the same as the first refiner 31 of the comparative example. The second refiner 32 of the present embodiment performs the refinement process in the same manner as the second refiner 32 of the comparative example, and outputs the same as the second refiner 32 of the comparative example.

  The above-described configurations of the motion vector compensation unit 90 and the object selection unit 20 will be described in more detail.

  The motion vector compensation unit 90 receives a signal mDir indicating a prediction direction for a certain block (that is, a coding target block). The motion vector compensation unit 90 receives the compensation mode cMode output from the mode determination unit 70. The motion vector compensation unit 90 receives the inherited motion vectors mMV0 and mMV1 generated by the pre-analysis unit 102. The motion vector compensation unit 90 further receives the above-described set A from the compensation candidate generation unit 60. The motion vector compensation unit 90 generates a compensation motion vector cMV based on the received mDir, cMode, mMV0, mMV1, and set A. The target selection unit 20 selects a motion vector used in the subsequent processing from mMV0, mMV1, and cMV. The target selection unit 20 stores (ie, stores) the selected motion vector (ie, the motion vector before refinement) separately in the first vector storage unit 81 and the second vector storage unit 82. Specifically, when the value of mDir is L0, the target selection unit 20 selects between mMV0 and cMV. The target selection unit 20 stores the mMV0 in the first vector storage unit 81 and stores the cMV in the second vector storage unit 82. When the value of mDir is L1, the target selection unit 20 selects cMV and mMV1. The target selection unit 20 stores the cMV in the first vector storage unit 81 and stores the mMV1 in the second vector storage unit 82. When the value of mDir is Bi, the target selection unit 20 selects mMV0 and mMV1. The target selection unit 20 stores mMV0 in the first vector storage unit 81 and stores mMV1 in the second vector storage unit 82. The target selection unit 20 may associate the motion vector before refinement with the current block to be coded (for example, the identifier of the current block to be coded). The target selection unit 20 may store the pre-refinement motion vector associated with the coding target block in the first vector storage unit 81 and the second vector storage unit 82 of the vector storage unit 80. The first vector storage unit 81 and the second vector storage unit 82 of the vector storage unit 80 only need to store the motion vector before refinement associated with the encoding target block.

  When the picture type (that is, the value of PicType) of the picture including the coding target block is a P picture, the target selection unit 20 of the present embodiment determines a vector used for generating a supplementary vector in another coding target block. You may choose. A vector used to generate a supplementary vector in another encoding target block is hereinafter referred to as a “propagation vector”. If the value of PicType is not a P picture, the target selection unit 20 does not need to select a propagation vector. The target selection unit 20 may receive the picture type PicType from the pre-analysis unit 102, for example. Note that the path of transmission of the picture type PicType from the pre-analysis unit 102 to the target selection unit 20 is not shown. When the picture type of the picture including the encoding target block is a P picture, for example, the inherited motion vector mMV0 and the supplementary vector cMV are input to the target selection unit 20.

  For example, the target selection unit 20 (for example, the first target selection unit 21) may select one of the inherited motion vector mMV0 and the compensation vector cMV as a propagation vector based on a predetermined propagation selection condition. The propagation condition is, for example, a condition based on the absolute value (that is, the magnitude) of the vector. The target selection unit 20 may select a vector having a large absolute value from the inherited motion vector mMV0 and the compensation vector cMV as the propagation vector. In this case, the above propagation selection condition is to have the largest absolute value. The target selection unit 20 may select a vector having a small absolute value from the inherited motion vector mMV0 and the compensation vector cMV as the propagation vector. In this case, the above-described propagation selection condition is to have the smallest absolute value.

  In this case, the target selection unit 20 stores the vector selected as the propagation vector from the motion vector mMV0 and the compensation vector cMV in the vector storage unit 80 so that it can be determined that the vector is selected as the propagation vector. do it. For example, the target selection unit 20 determines whether the target selection unit 20 determines whether the vector selected as the propagation vector among the motion vector mMV0 and the compensation vector cMV is a propagation vector (hereinafter, “determination”). Value). Then, the target selection unit 20 may store, in the vector storage unit 80, the motion vector mMV0 and the compensation vector cMV, each of which is associated with a determination value.

  The target selection unit 20 may store the vector selected as the propagation vector in a fixed one of the first vector storage unit 81 and the second vector storage unit 82 included in the vector storage unit 80. In this case, the target selection unit 20 may store the vector that has not been selected as the propagation vector. Further, the target selection unit 20 may store the target in the other of the first vector storage unit 81 and the second vector storage unit 82. In the following description, the target selection unit 20 stores the vector selected as the propagation vector in the first vector storage unit 81. Further, the target selection unit 20 stores the vector not selected as the propagation vector in the second vector storage unit 82. For example, when the inherited motion vector mMV0 is selected as the propagation vector from the inherited motion vector mMV0 and the compensation vector cMV, the target selection unit 20 (for example, the first target selection unit 21) replaces the inherited motion vector mMV0 with the first vector. It is stored in the storage unit 81. In this case, the target selection unit 20 (for example, the second target selection unit 22) stores the compensation vector cMV in the second vector storage unit 82. When the compensation vector cMV is selected as the propagation vector from the inherited motion vector mMV0 and the compensation vector cMV, the target selection unit 20 (for example, the second target selection unit 22) stores the compensation vector cMV in the first vector storage unit 81. Store. In this case, the target selection unit 20 (for example, the first target selection unit 21) stores the inherited motion vector mMV0 in the second vector storage unit 82.

  The compensation candidate generation unit 60 may read the vector stored in the vector storage unit 80 as the compensation vector as the peripheral vector used to generate the AMVP motion vector and the temporal prediction motion vector.

  Next, the motion vector compensation unit 90 will be described in more detail.

  FIG. 13 is a block diagram illustrating an example of the configuration of the motion vector compensation unit 90 according to the present embodiment in detail. The motion vector compensation unit 90 includes a first candidate selection unit 95, a second candidate selection unit 96, a selection execution unit 97, a third candidate selection unit 98, and a propagation condition determination unit 99. The first candidate selecting unit 95 is the same as the first candidate selecting unit 15 shown in FIG. The second candidate selection unit 96 is the same as the second candidate selection unit 16 shown in FIG. Further, the third candidate selection unit 98 receives as input all the predicted motion vectors (that is, the vectors included in the set A, also referred to as “candidate vectors”) input from the compensation candidate generation unit 60. The vectors included in set A are AMVP00, AMVP01, AMVP10, AMVP11, temporally predicted motion vector 0, temporally predicted motion vector 1, and zero vector, as described above. The third candidate selection unit 98 receives the propagation condition from the propagation condition determination unit 99. The propagation conditions will be described later in detail. The third candidate selection unit 98 selects one vector from the received vectors (that is, the vectors included in the set A) based on the received propagation conditions. The third candidate selection unit 98 outputs the selected one vector as the pseudo propagation compensation vector cMVP0 and the pseudo propagation compensation vector cMVP1. That is, the third candidate selection unit 98 transmits the selected one vector to the selection execution unit 97 as the pseudo propagation compensation vector cMVP0 and the pseudo propagation compensation vector cMVP1. The pseudo propagation compensation vector cMVP0 is a candidate for a compensation vector for L0 (that is, a candidate for a compensation vector to be compensated when mMV0 does not exist). The pseudo propagation compensation vector cMVP1 is a candidate for a compensation vector for L1.

  The propagation condition determination unit 99 determines, for example, one of the following propagation conditions as a propagation condition to be transmitted to the third candidate selection unit 98. Then, the propagation condition determining unit 99 outputs (ie, transmits) the determined propagation condition to the third candidate selecting unit 98.

(A) The vector having the smallest absolute value (that is, the magnitude) of the vector among the plurality of motion vectors. (B) The vector (C) having the second smallest magnitude (that is, the magnitude) of the vector among the plurality of motion vectors. ) Among the plurality of motion vectors, the vector having the second largest absolute value (ie, magnitude) of the vector. (D) Of the plurality of motion vectors, the vector having the largest absolute value (ie, magnitude) of the vector. Of the motion vectors, the vector having the smallest distance to the inherited motion vector. (F) Of the plurality of motion vectors, the vector having the second smallest distance to the inherited motion vector. (G) Of the plurality of motion vectors, the inherited motion vector. The vector having the second largest distance to the vector (H) Among the plurality of motion vectors, the vector having the largest distance to the inherited motion vector (I) Among the motion vectors, the vector whose absolute value is smaller than the threshold is the vector having the largest distance to the inherited motion vector. (J) Among the plurality of motion vectors, the vector whose absolute value is larger than the threshold is The vector whose distance to the motion vector is the smallest (K) Among the plurality of motion vectors, the vector whose absolute value is the largest among the vectors whose distance to the inherited motion vector is smaller than the threshold value (L) Of the plurality of motion vectors , The vector having the smallest absolute value among the vectors having a distance from the inherited motion vector larger than the threshold value. The propagation condition determining unit 99 may fixedly output a propagation condition experimentally determined in advance. . Further, the propagation condition determination unit 99 may dynamically determine the propagation condition by using encoding information (not shown) at the time of executing the encoding.

  The selection execution unit 97 receives the compensation vector cMV0 from the first candidate selection unit 95. The selection execution unit 97 receives the supplementary vector cMV1 from the second candidate selection unit 96. The selection execution unit 97 receives the pseudo propagation compensation vectors cMVP0 and cMVP1 from the third candidate selection unit 98. The selection executing unit 97 selects one of the maximum four of cMV0, cMV1, cMVP0, and cMVP1 (hereinafter also referred to as “secondary candidate vector”) as cMV based on mDir. The selection execution unit 97 may further receive the supplement mode cMode. The selection execution unit 97 may select one vector as a cMV from among the maximum four vectors based on mDir and cMode (that is, may select the cMV). The selection execution unit 97 may further receive the picture type PicType. The selection execution unit 97 may further select a cMV based on PicType. The selection execution unit 97 outputs the selected vector as cMV. Various methods are applicable as a method for the selection execution unit 97 to select the cMV. The method by which the selection execution unit 97 selects the cMV may be determined as appropriate.

  For example, when the value of mDir is L0 (that is, when mMV0 exists and mMV1 does not exist), the selection execution unit 97 may select one of cMV0 and cMVP1 according to a predetermined selection method. . When the value of mDir is L1 (that is, when mMV1 exists and mMV0 does not exist), the selection execution unit 97 may select either cMV1 or cMVP0 according to a predetermined selection method.

  The selection execution unit 97 may select, for example, a vector having the smallest absolute value as a cMV among a plurality of vectors to be selected. For example, as described above, when selecting one of the two vectors as the cMV, the selection execution unit 97 may select the vector having the smaller absolute value as the cMV from the two vectors. In the above example, when the value of mDir is L0, the selection execution unit 97 may select a vector having a small absolute value from cMV0 and cMVP1 as cMV. When the value of mDir is L1, the selection execution unit 97 may select a vector having a small absolute value from cMV1 and cMVP0 as cMV. The selection executing unit 97 may select the compensation vector cMV0 or cMV1 as the compensation vector cMV based on the difference between the absolute values of the compensation vector cMV0 or cMV1 and the pseudo propagation compensation vectors cMVP0 and cMVP1. For example, when the value of the prediction direction mDir is L0, if the difference between the absolute value of cMV0 and the absolute value of cMVP1 is equal to or larger than the threshold, the selection execution unit 97 may select cMVP1. When the value of the prediction direction mDir is L1, if the difference between the absolute value of cMV1 and the absolute value of cMVP0 is equal to or greater than the threshold, the selection execution unit 97 may select cMVP0.

  The selection executing unit 97 may select the cMV according to another method.

  FIG. 14 is a block diagram schematically illustrating an example of the configuration of the motion vector compensation section 90 of the present embodiment.

  The motion vector compensation unit 90 shown in FIG. 13 can be drawn in a simplified configuration as shown in FIG. The motion vector compensation unit 90 illustrated in FIG. 14 includes a detection unit 91, a compensation vector selection unit 92, a reception unit 94, and a transmission unit 93. The detection unit 91 detects the prediction direction mDir (specifically, the value of the prediction direction mDir) transmitted from the pre-analysis unit 102. The receiving unit 94 receives the inherited motion vector (at least one of mMV0 and mMV1) from the pre-analysis unit 102. The receiving unit 94 receives, for example, the compensation mode cMode from the mode determining unit 70. The receiving unit 94 further receives the vector of the set A from the compensation candidate generating unit 60. The compensation vector selection unit 92 compensates by selecting one vector from the vectors of the set A as the compensation vector cMV based on the prediction direction mDir and the inherited motion vector according to the propagation condition determined by the propagation condition determination unit 99. Generate a vector cMV. The transmission unit 93 transmits the generated compensation vector cMV to the target selection unit 20.

  As in the comparative example described above, the part corresponding to the receiving unit 94 is a part of the motion vector compensation unit 90 that receives the inherited motion vector, the compensation mode cMode, and the set A. The part corresponding to the detection unit 91 is a part of the motion vector compensation unit 90 that receives the prediction direction mDir and the picture type PicTYpe and determines the values of the received prediction direction mDir and the picture type PicTYpe (for example, the selection execution unit). 97 part). Further, the first candidate selection unit 95, the second candidate selection unit 96, and the selection execution unit 97 operate as the compensation vector selection unit 92. Further, a part corresponding to the transmission unit 93 is a part that transmits the cMV from the selection execution unit 97 to the target selection unit 20.

[Description of operation]
Next, the operation of the motion compensation unit 200 of the video encoding device 1 according to the present embodiment will be described in detail with reference to the drawings. Note that the entire operation of the video encoding device 1 according to the present embodiment is the same as the operation of a general video encoding device, and thus the description thereof will be omitted.

  FIGS. 15 and 16 are flowcharts illustrating the operation of the motion compensation unit 200 according to the present embodiment. At the start of the operation illustrated in FIG. 15, the video is divided into, for example, the largest coding unit (Largest Coding Unit, hereinafter referred to as “LCU”), which is a group of blocks to be encoded at one time, by the block dividing unit 101. ing. The maximum coding unit includes a plurality of prediction units (Prediction Unit, hereinafter referred to as “PU”). The PU is a block that is a unit of prediction in video encoding (that is, the smallest block from which a prediction vector, a prediction image, and a prediction direction are derived). Then, the motion compensation unit 200 performs the operation shown in FIGS. 15 and 16 for each LCU. At the start of the operation shown in FIG. 15, the LCU to be processed has been selected.

  The motion compensation unit 200 first performs the operations of Step S101, Step S102, and Steps S302 to S305 for each PU of the selected LCU (Step S301).

  Specifically, first, the mode determination unit 70 determines a compensation mode, and transmits the decided compensation mode to the motion vector compensation unit 90 (step S101). As in the operation of step S101 in the comparative example shown in FIG. 8, when the compensation mode is fixed, the mode determination unit 70 does not need to determine the compensation mode.

  Next, the compensation candidate generating unit 60 generates a vector of the set A including the AMVP predicted motion vector and the temporal predicted motion vector based on the surrounding vectors (step S102). In the present embodiment, the peripheral vector is a motion vector before refinement of a block adjacent to the current block (that is, the above-described PU) or a motion vector before refinement of a block of a picture immediately before the current block. As described above, the vector storage unit 80 stores the peripheral vector (that is, the motion vector before refinement in a block around the encoding target block). The compensation candidate generation unit 60 may read the peripheral vectors from the vector storage unit 80, and generate the vector of the set A based on the read peripheral vectors.

  Next, the motion vector compensation unit 90 performs a motion vector compensation process (step S302). The motion vector compensation processing will be described later in detail. By the motion vector compensation processing, two inherited motion vectors or a combination of the inherited motion vector and the compensated motion vector is obtained as the motion vector before refinement.

  If the picture type of the picture including the coding target block is a P picture (YES in step S303), the target selection unit 20 selects one of the two pre-refinement motion vectors as the propagation vector (step S304). If the picture type of the picture including the encoding target block is not a P picture (NO in step S303), the target selection unit 20 need not select a propagation vector.

  Next, the target selection unit 20 stores the motion vector obtained as the pre-refinement motion vector in the vector storage unit 80 (Step S305). The target selection unit 20 may associate the coding target block (that is, the above-described PU) with the motion vector obtained as the motion vector before refinement. Then, the target selection unit 20 may store the motion vector associated with the coding target block in the vector storage unit 80. The target selection unit 20 may associate the motion vector stored in the vector storage unit 80 with, for example, the number specifying the PU described above.

  When the picture type of the picture including the encoding target block is a P picture, the target selection unit 20 stores the two pre-refinement motion vectors in the vector storage unit 80 so as to specify which vector has been selected as the propagation vector. Just store it. In the example of the operation of the present embodiment, the target selection unit 20 stores the vector selected as the propagation vector in the first vector storage unit 81. Then, the target selection unit 20 stores the vector not selected as the propagation vector in the second vector storage unit 82.

  The motion compensation unit 200 repeats the operations of step S101, step S102, step S302, and step S303 for each PU of the selected LCU until the motion vector compensation processing ends (step S304).

  Next, the motion compensation unit 200 performs the operation shown in FIG. The motion compensation unit 200 may perform the operations from step S104 to step S109 for each PU in the selected LCU (step S311). The operation from step S104 to step S109 is the same as the operation from step S104 to step S109 of the motion compensation unit 100 of the modified example shown in FIG. The motion compensation unit 200 repeats the operation from step S104 to step S109 for each PU in the selected LCU until the motion vector, the predicted image, and the prediction direction are output (step S312).

  Next, the motion vector compensation processing by the motion vector compensation unit 90 will be described in detail with reference to the drawings.

  FIG. 17 is a flowchart illustrating an example of the operation of the motion vector compensation processing by the motion vector compensation unit 90 of the present embodiment.

  Referring to FIG. 17, first, the detection unit 91 of the motion vector compensation unit 90 shown in FIG. 14 detects the number of inherited motion vectors input to the motion compensation unit 200 based on the prediction direction mDir (step S201). ). Specifically, for example, the selection execution unit 97 (operating as the above-described detection unit 91) illustrated in FIG. 13 may detect the number of inherited motion vectors input to the motion compensation unit 200. The detecting unit 91 may determine that the number of inherited motion vectors is 1 when the value of the prediction direction mDir is L0 or L1, as in the detecting unit 11 of the comparative example. When the value of the prediction direction mDir is Bi, the detection unit 91 may determine that the number of inherited motion vectors is two. When the number of inherited motion vectors is not 1 (NO in step S202), the motion vector compensation unit 90 ends the operation illustrated in FIG.

  If the number of inherited motion vectors is 1 (YES in step S202), the motion vector compensation unit 90 (for example, a part operating as the reception unit 94) receives the compensation mode cMode from, for example, the mode determination unit 70 (step S203). ). Specifically, the first candidate selection unit 95 and the second candidate selection unit 96 receive the compensation mode cMode. The third candidate selection unit 98 may also receive the compensation mode cMode.

  The motion vector compensation unit 90 (for example, a part that operates as the reception unit 94) further receives the vector of the set A from the compensation candidate generation unit 60 (Step S204). Specifically, the first candidate selection unit 95 receives the vector of the set 0 among the vectors of the set A. The second candidate selection unit 96 receives the vector of the set 1 among the vectors of the set A. The third candidate selection unit 98 receives the vector of the set A. As described above, the vectors of the set A are AMVP00, AMVP01, AMVP10, AMVP11, temporally predicted motion vector 0, temporally predicted motion vector 1, and zero vector. The vectors of set 0 are AMVP00, AMVP01, temporal prediction motion vector 0, and zero vector. The vectors of set 1 are AMVP10, AMVP11, temporal motion vector predictor 1, and zero vector.

  The motion vector compensation unit 90 (for example, a part operating as the receiving unit 94) further receives the inherited motion vector from the pre-analysis unit 102 (Step S205). Specifically, the first candidate selection unit 95 receives the inherited motion vector mMV0. The second candidate selection unit 96 receives the inherited motion vector mMV1. The second candidate selection unit 96 receives the inherited motion vector mMV0 and the inherited motion vector mMV1. When the pre-analysis unit 102 does not generate the inherited motion vector mMV0, the first candidate selecting unit 95 and the third candidate selecting unit 98 need not receive the inherited motion vector mMV0. When the pre-analysis unit 102 does not generate the inherited motion vector mMV1, the second candidate selecting unit 96 and the third candidate selecting unit 98 need not receive the inherited motion vector mMV1.

  Next, the motion vector compensation unit 90 generates at least one of the compensation vectors cMV0 and cMV1 based on the compensation mode cMode (step S401). Specifically, the first candidate selection unit 95 selects one vector from the vectors of the set 0 according to the method defined for the received compensation mode cMode. The first candidate selection unit 95 may select a vector using the inherited motion vector mMV0. Then, the first candidate selecting unit 95 generates the compensation vector cMV0 by setting the selected vector to the compensation vector cMV0. The first candidate selection unit 95 transmits the generated compensation vector cMV0 to the selection execution unit 97. The second candidate selecting unit 96 selects one vector from the vectors of the set 1 in accordance with the method defined for the received compensation mode cMode. The second candidate selection unit 96 may select a vector using the inherited motion vector mMV1. Then, the second candidate selecting unit 96 generates the compensation vector cMV1 by setting the selected vector as the compensation vector cMV1. The second candidate selection unit 96 transmits the generated compensation vector cMV1 to the selection execution unit 97.

  Next, the propagation condition determining unit 99 determines a propagation condition (Step S402). The propagation condition determination unit 99 may determine the propagation condition based on the encoding state. The propagation condition may be fixed. In that case, the propagation condition determination unit 99 need not determine the propagation condition. The propagation condition determining unit 99 transmits the propagation condition to the third candidate selecting unit 98. Specifically, for example, the propagation condition determining unit 99 may transmit, to the third candidate selecting unit 98, an identifier indicating a transmission condition to be transmitted, among identifiers predetermined for each of the plurality of propagation conditions. . The third candidate selection unit 98 receives the propagation condition from the propagation condition determination unit 99. Specifically, the third candidate selecting unit 98 may receive the identifier of the propagation condition.

  Next, the motion vector compensation unit 90 generates a pseudo propagation compensation vector based on the vectors included in the set A according to the received propagation condition (step S403). Specifically, the third candidate selecting unit 98 selects a vector satisfying the received propagation condition (that is, a vector satisfying the propagation condition indicated by the received identifier) from the vectors included in the set A. The third candidate selection unit 98 generates pseudo propagation compensation vectors cMVP0 and cMVP1 by setting the selected vectors as pseudo propagation compensation vectors cMVP0 and cMVP1. The third candidate selection unit 98 transmits the pseudo propagation compensation vectors cMVP0 and cMVP1 to the selection execution unit 97.

  Next, the motion vector compensation unit 90 selects a compensation vector cMV from the compensation vectors cMV0 and cMV1 and the pseudo propagation compensation vectors cMVP0 and cMVP1 according to a predetermined propagation selection condition (step S404). The propagation selection condition is, for example, to have the largest absolute value. In this case, specifically, when the prediction direction mDir is L0, the selection execution unit 97 selects a vector having a large absolute value from, for example, the compensation vector cMV0 and the pseudo propagation compensation vector cMVP1. When the prediction direction mDir is L1, the selection execution unit 97 selects, for example, a vector having a large absolute value from the compensation vector cMV1 and the pseudo propagation compensation vector cMVP0. The propagation selection condition may be, for example, to provide the smallest absolute value. In this case, when the prediction direction mDir is L0, the selection execution unit 97 selects a vector having a small absolute value from, for example, the compensation vector cMV0 and the pseudo propagation compensation vector cMVP1. When the prediction direction mDir is L1, the selection execution unit 97 selects, for example, a vector having a small absolute value from the compensation vector cMV1 and the pseudo propagation compensation vector cMVP0.

  Note that the order of the operations from step S203 to step S403 is not limited to the example shown in FIG. For example, the operations in steps S203 to S205 may be performed before the operations in steps S401 and S403. The operation in step S402 may be performed before the operation in step S403. The order in which the operations from step S203 to step S205 and the operation in step S402 are performed can be appropriately changed. At least two of the operations from step S203 to step S205 and the operation in step S402 may be performed in parallel. The order in which the operation in step S401 and the operation in step S403 are performed may be reversed. The operation in step S401 and the operation in step S403 may be performed in parallel.

  Then, the motion vector compensation unit 90 ends the operation illustrated in FIG.

  FIG. 18 is a diagram schematically illustrating a portion that performs motion vector compensation and refinement in the video encoding device 2 of the present embodiment. As in FIG. 7, mcMV0 and mcMV1 shown in FIG. 18 represent two pre-refined motion vectors selected by the target selection unit 20 among the mMV0, mMV1, and cMV. McMV0a and mcMV1a shown in FIG. 18 represent a motion vector before refinement of a peripheral block of the current block.

[Explanation of effects]
Next, effects of the present embodiment will be described.

  The present embodiment has an effect that the prediction accuracy can be improved even when the number of reference pictures is one.

  The reason is that the motion vector compensation unit 90 selects a compensation vector as a refinement target vector from a combination of vectors including the AMVP prediction vector and the temporal prediction vector. If there is one reference picture, there is one inherited motion vector. In that case, two prediction vectors can be generated by refining the inherited motion vector and the selected compensation vector. By performing motion compensation based on two prediction vectors, the accuracy of prediction is improved. Thereby, the compression ratio of the video is improved.

  Further, the motion vector compensation unit 90 selects a compensation vector from the AMVP prediction motion vector selected from the inherited motion vector and the compensation vector in the peripheral block. As a result, propagation of a vector that occurs in ideal vector filling can be realized in a pseudo manner. Thereby, the accuracy of the prediction is further improved.

[Second embodiment]
FIG. 19 is a block diagram illustrating an example of a configuration of a video encoding device according to the second embodiment of the present invention.

  Referring to FIG. 19, a video encoding device 2A according to the second embodiment of the present invention includes a compensation candidate generation unit 60 and a motion vector compensation unit 90.

  The compensation candidate generation unit 60 generates a candidate vector including the pre-refinement vector of the block adjacent to the encoding target block. The pre-refinement vector is a vector selected as a criterion for searching for a motion vector predictor that indicates motion prediction of a block into which a picture included in the video to be encoded is divided. The candidate vector is, for example, the vector of the set A in the above-described first embodiment.

  The motion vector compensation unit 90 selects, from the candidate vectors, the compensation vector compensated as the pre-refinement vector when the number of inherited motion vectors obtained as the pre-refinement vector of the current block is one. I do. The compensation vector is the compensation vector cMV of the above-described first embodiment. The method of selecting the compensation vector may be the same as the method of selecting the compensation vector cMV in the first embodiment.

  Next, the operation of the video encoding device 2A of the present embodiment will be described in detail with reference to the drawings.

  FIG. 20 is a flowchart illustrating an example of the operation of the video encoding device according to the present embodiment.

Referring to FIG. 20, first, the compensation candidate generating unit 60 generates a candidate vector including a pre-refinement vector of an adjacent block (step S501). Next, the motion vector compensation unit 90 selects a compensation vector to be supplemented as a pre-refinement vector of the encoding target block from the candidate vectors (step S502).
[Other embodiments]
The video encoding devices according to the above-described embodiment and the comparative example can be realized by a computer and a program for controlling the computer, dedicated hardware, or a combination of the computer and the program for controlling the computer and dedicated hardware, respectively. Can be.

  In other words, the video encoding device according to each of the above-described embodiments and the comparative example can be realized by hardware such as a circuit configuration. The circuit configuration may be, for example, a processor and a memory included in the computer. In that case, the program only needs to be loaded in the memory. The program can be executed by a processor, and the computer may be operated as the video encoding device of each of the above-described embodiments and the comparative example. The circuit configuration may be, for example, a plurality of computers communicably connected. The circuit configuration may be, for example, a circuit. The circuit configuration may be, for example, a plurality of circuits communicably connected. The circuit configuration may be a combination of one or more computers and one or more circuits communicably connected.

  FIG. 21 is a diagram illustrating an example of a hardware configuration of a computer 1000 that can realize the video encoding devices according to the above-described embodiment and the comparative example. Referring to FIG. 21, the computer 1000 includes a processor 1001, a memory 1002, a storage device 1003, and an I / O (Input / Output) interface 1004. In addition, the computer 1000 can access the recording medium 1005. The memory 1002 and the storage device 1003 are storage devices such as a RAM (Random Access Memory) and a hard disk. The recording medium 1005 is, for example, a storage device such as a RAM or a hard disk, a ROM (Read Only Memory), or a portable recording medium. The storage device 1003 may be the recording medium 1005. The processor 1001 can read and write data and programs from and to the memory 1002 and the storage device 1003. The processor 1001 can access, for example, a device that supplies a video signal and a device to which an encoded video signal is supplied, via the I / O interface 1004. The processor 1001 can access the recording medium 1005. The recording medium 1005 stores a program that causes the computer 1000 to operate as the video encoding device 1, 2, or 2A.

  The processor 1001 loads a program stored in the recording medium 1005 for operating the computer 1000 as the video encoding device 1, 2, or 2A into the memory 1002. When the processor 1001 executes the program loaded in the memory 1002, the computer 1000 operates as the video encoding device 1, 2, or 2A.

  The following units (hereinafter, referred to as “first group”) are realized by, for example, a memory 1002 loaded with a dedicated program capable of realizing the functions of those units, and a processor 1001 executing the program. be able to.

Motion vector compensation unit 10
Detector 11
Compensation vector selection unit 12
Transmission unit 13
Receiver 14
First candidate selection unit 15
Second candidate selection unit 16
Selection execution unit 17
Object selection unit 20
First object selection unit 21
Second object selection unit 22
Refine part 30
First refiner 31
Second refiner 32
Bi-prediction generation unit 40
Prediction selection unit 50
Compensation candidate generation unit 60
Mode determination unit 70
Motion vector compensation unit 90
Detector 91
Compensation vector selection unit 92
Transmission section 93
Receiver 94
First candidate selection unit 95
Second candidate selection unit 96
Selection execution unit 97
Third candidate selection unit 98
Propagation condition determination unit 99
Motion compensation unit 100
Block dividing unit 101
Pre-analysis unit 102
Intra prediction unit 103
Switching unit 104
Integer conversion unit 105
Quantization unit 106
Inverse quantization unit 107
Inverse integer converter 108
Encoding section 109
Filter section 116
Subtraction unit 118
Adder 119
Motion compensation unit 200
Further, the following units (hereinafter, referred to as “second group”) can be realized by a memory 1002 included in the computer 1000 or a storage device 1003 such as a hard disk device.

Vector storage unit 80
First vector storage unit 81
Second vector storage unit 82
Frame storage unit 117
Some or all of the units included in the first group and the units included in the second group may be realized by a dedicated circuit that realizes the function of each unit.

  FIG. 22 is a block diagram illustrating an example of a configuration of a video encoding device 1 implemented by a circuit according to a comparative example of the present invention. The video encoding device 1 illustrated in FIG. 22 includes a block division circuit 1101, a pre-analysis circuit 1102, an intra prediction circuit 1103, a switching circuit 1104, an integer conversion circuit 1105, a quantization circuit 1106, an inverse quantization circuit 1107, and an inverse integer. A conversion circuit 1108 is included. The video encoding device 1 further includes an encoding circuit 1109, a filter circuit 1116, a frame storage circuit 1117, a subtraction circuit 1118, an addition circuit 1119, and a motion compensation circuit 1100.

  FIG. 23 is a block diagram illustrating an example of a configuration of a motion compensation circuit 1100 according to a comparative example of the present invention. The motion compensation circuit 1100 shown in FIG. 23 includes a motion vector compensation circuit 1010, an object selection circuit 1020, a refinement circuit 1030, a bi-prediction generation circuit 1040, a prediction selection circuit 1050, a compensation candidate generation circuit 1060, and a mode determination circuit 1070. .

  FIG. 24 is a block diagram illustrating an example of a configuration of a motion compensation circuit 1100 according to a comparative example of the present invention. FIG. 24 shows an example of a more detailed configuration of the motion compensation circuit 1100 shown in FIG. The motion compensation circuit 1100 shown in FIG. 24 includes a motion vector compensation circuit 1010, an object selection circuit 1020, a refinement circuit 1030, a bi-prediction generation circuit 1040, a prediction selection circuit 1050, a compensation candidate generation circuit 1060, and a mode determination circuit 1070. . The target selection circuit 1020 includes a first target selection circuit 1021 and a second target selection circuit 1022. The refine circuit 1030 includes a first refine circuit 1031 and a second refine circuit 1032.

  FIG. 25 is a block diagram illustrating an example of a configuration of a motion vector compensation circuit 1010 according to a comparative example of the present invention. The motion vector compensation circuit 1010 shown in FIG. 25 includes a first candidate selection circuit 1015, a second candidate selection circuit 1016, and a selection execution circuit 1017.

  FIG. 26 is a block diagram illustrating an example of a configuration of a motion vector compensation circuit 1010 according to a comparative example of the present invention. The motion vector compensation circuit 1010 shown in FIG. 25 can also be drawn as shown in FIG. The motion vector compensation circuit 1010 shown in FIG. 26 includes a detection circuit 1011, a compensation vector selection circuit 1012, a transmission circuit 1013, and a reception circuit 1014.

  FIG. 27 is a block diagram illustrating an example of a configuration of a video encoding device 2 implemented by a circuit according to the third embodiment of this invention. The video encoding device 2 illustrated in FIG. 27 includes a block division circuit 1101, a pre-analysis circuit 1102, an intra prediction circuit 1103, a switching circuit 1104, an integer conversion circuit 1105, a quantization circuit 1106, an inverse quantization circuit 1107, and an inverse integer. A conversion circuit 1108 is included. The video encoding device 2 further includes an encoding circuit 1109, a filter circuit 1116, a frame storage circuit 1117, a subtraction circuit 1118, an addition circuit 1119, and a motion compensation circuit 1200.

  FIG. 28 is a block diagram illustrating an example of a configuration of the motion compensation circuit 1200 according to the first embodiment of this invention. The motion compensation circuit 1200 shown in FIG. 28 includes a motion vector compensation circuit 1090, an object selection circuit 1020, a refinement circuit 1030, a bi-prediction generation circuit 1040, a prediction selection circuit 1050, a compensation candidate generation circuit 1060, a mode determination circuit 1070, and a vector A storage circuit 1080 is included.

  FIG. 29 is a block diagram illustrating an example of a configuration of a motion compensation circuit 1200 implemented by a circuit according to the first embodiment of this invention. FIG. 29 shows an example of a more detailed configuration of the motion compensation circuit 1100 shown in FIG. The motion compensation circuit 1200 shown in FIG. 29 includes a motion vector compensation circuit 1090, an object selection circuit 1020, a refinement circuit 1030, a bi-prediction generation circuit 1040, a prediction selection circuit 1050, a compensation candidate generation circuit 1060, a mode determination circuit 1070, and a vector A storage circuit 1080 is included. The target selection circuit 1020 includes a first target selection circuit 1021 and a second target selection circuit 1022. The refine circuit 1030 includes a first refine circuit 1031 and a second refine circuit 1032. The vector storage circuit 1080 includes a first vector storage circuit 1081 and a second vector storage circuit 1082.

  FIG. 30 is a block diagram illustrating an example of a configuration of the motion vector compensation circuit 1090 according to the first embodiment of this invention. The motion vector compensation circuit 1090 shown in FIG. 30 includes a first candidate selection circuit 1095, a second candidate selection circuit 1096, a third candidate selection circuit 1098, a selection execution circuit 1097, and a propagation condition determination circuit 1099.

  FIG. 31 is a block diagram illustrating an example of a configuration of the motion vector compensation circuit 1090 according to the first embodiment of this invention. The motion vector compensation circuit 1090 shown in FIG. 30 can be drawn as shown in FIG. The motion vector compensation circuit 1090 shown in FIG. 31 includes a detection circuit 1091, a compensation vector selection circuit 1092, a transmission circuit 1093, a reception circuit 1094, and a propagation condition determination circuit 1099.

  FIG. 32 is a block diagram illustrating an example of a configuration of a video encoding device 2A implemented by a circuit according to the second embodiment of this invention. The video encoding device 2A shown in FIG. 32 includes a compensation candidate generation circuit 1060 and a motion vector compensation circuit 1090.

  22 to 32, the motion vector compensation circuit 1010 operates as the motion vector compensation unit 10. The detection circuit 1011 operates as the detection unit 11. The compensation vector selection circuit 1012 operates as the compensation vector selection unit 12. The transmission circuit 1013 operates as the transmission unit 13. The receiving circuit 1014 operates as the receiving unit 14. The first candidate selection circuit 1015 operates as the first candidate selection unit 15. The second candidate selection circuit 1016 operates as the second candidate selection unit 16. The selection execution circuit 1017 operates as the selection execution unit 17. The target selection circuit 1020 operates as the target selection unit 20. The first target selection circuit 1021 operates as the first target selection unit 21. The second target selection circuit 1022 operates as the second target selection unit 22. The refine circuit 1030 operates as the refine unit 30. The first refine circuit 1031 operates as the first refine unit 31. The second refiner 1032 operates as the second refiner 32. The bi-prediction generation circuit 1040 operates as the bi-prediction generation unit 40. The prediction selection circuit 1050 operates as the prediction selection unit 50. The compensation candidate generation circuit 1060 operates as the compensation candidate generation unit 60. The mode determining circuit 1070 operates as the mode determining unit 70.

  The vector storage circuit 1080 operates as the vector storage unit 80. The first vector storage circuit 1081 operates as the first vector storage unit 81. The second vector storage circuit 1082 operates as the second vector storage unit 82. The motion vector compensation circuit 1090 operates as the motion vector compensation unit 90. The detection circuit 1091 operates as the detection unit 91. The compensation vector selection circuit 1092 operates as the compensation vector selection unit 92. The transmission circuit 1093 operates as the transmission unit 93. The receiving circuit 1094 operates as the receiving unit 94. The first candidate selection circuit 1095 operates as the first candidate selection unit 95. The second candidate selection circuit 1096 operates as the second candidate selection unit 96. The selection execution circuit 1097 operates as the selection execution unit 97. The third candidate selection circuit 1098 operates as the third candidate selection unit 98. The propagation condition determination circuit 1099 operates as the propagation condition determination unit 99.

  The motion compensation circuit 1100 operates as the motion compensation unit 100. The block division circuit 1101 operates as the block division unit 101. The pre-analysis circuit 1102 operates as the pre-analysis unit 102. The intra prediction circuit 1103 operates as the intra prediction circuit 103. The switching circuit 1104 operates as the switching unit 104. The integer conversion circuit 1105 operates as the integer conversion unit 105. The quantization circuit 1106 operates as the quantization unit 106. The inverse quantization circuit 1107 operates as the inverse quantization unit 107. The inverse integer conversion circuit 1108 operates as the inverse integer conversion unit 108. The encoding circuit 1109 operates as the encoding unit 109. The filter circuit 1116 operates as the filter unit 116. The frame storage circuit 1117 operates as the frame storage unit 117. The subtraction circuit 1118 operates as the subtraction unit 118. The adding circuit 1119 operates as the adding unit 119. The motion compensation circuit 1200 operates as the motion compensation unit 200.

  Further, some or all of the above-described embodiments can be described as in the following supplementary notes, but are not limited to the following.

(Appendix 1)
Among the pre-refinement vectors selected as a reference for searching for a predicted motion vector representing the prediction of the motion of the block into which the picture included in the video to be encoded is divided, the pre-refinement vector of the block adjacent to the current block. Supplementation candidate generating means for generating a candidate vector, comprising:
A motion vector compensation unit for selecting, from the candidate vectors, the compensation vector compensated as the pre-refinement vector when the number of inherited motion vectors obtained as the pre-refinement vector of the encoding target block is 1,
A video encoding device comprising:

(Appendix 2)
The video encoding device according to claim 1, wherein the motion vector compensation unit selects the compensation vector based on a distance between the inherited motion vector and the candidate vector.

(Appendix 3)
The video encoding device according to Supplementary Note 1 or 2, wherein the motion vector compensation unit selects the compensation vector based on a size of the candidate vector.

(Appendix 4)
4. The motion vector compensation unit according to claim 1, wherein the motion vector compensation unit selects a plurality of secondary candidate vectors according to different conditions from the candidate vectors, and selects the compensation vector from the selected plurality of secondary candidate vectors. 5. Video encoding device.

(Appendix 5)
Among the pre-refinement vectors selected as a reference for searching for a predicted motion vector representing the prediction of the motion of the block into which the picture included in the video to be encoded is divided, the pre-refinement vector of the block adjacent to the current block. Generate a candidate vector containing
When the number of inherited motion vectors obtained as the pre-refinement vector of the encoding target block is 1, the compensation vector to be compensated as the pre-refinement vector is selected from the candidate vectors,
Video encoding method.

(Appendix 6)
The video encoding method according to Supplementary Note 5, wherein the compensation vector is selected based on a distance between the inherited motion vector and the candidate vector.

(Appendix 7)
7. The video encoding method according to claim 5, wherein the compensation vector is selected based on a size of the candidate vector.

(Appendix 8)
The video encoding method according to any one of Supplementary Notes 5 to 8, wherein a plurality of secondary candidate vectors are selected according to different conditions from the candidate vectors, and the supplementary vector is selected from the selected secondary candidate vectors.

(Appendix 9)
On the computer,
Among the pre-refinement vectors selected as a reference for searching for a predicted motion vector representing the prediction of the motion of the block into which the picture included in the video to be encoded is divided, the pre-refinement vector of the block adjacent to the current block. A compensation candidate generation process for generating a candidate vector, including:
When the number of inherited motion vectors obtained as the pre-refinement vector of the encoding target block is 1, the compensation vector to be supplemented as the pre-refinement vector, a motion vector compensation process of selecting from the candidate vectors,
A video encoding program that executes

(Appendix 10)
The video encoding program according to supplementary note 9, wherein the motion vector compensation processing selects the compensation vector based on a distance between the inherited motion vector and the candidate vector.

(Appendix 11)
The video encoding program according to attachment 9 or 10, wherein the motion vector compensation processing selects the compensation vector based on a size of the candidate vector.

(Appendix 12)
12. The motion vector compensation process according to claim 9, further comprising selecting a plurality of secondary candidate vectors according to different conditions from the candidate vectors, and selecting the compensation vector from the selected plurality of secondary candidate vectors. Video coding program.

  As described above, the present invention has been described with reference to the exemplary embodiments. However, the present invention is not limited to the exemplary embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

  ADVANTAGE OF THE INVENTION According to this invention, it is applicable to uses, such as a video compression apparatus which compresses an image by encoding, and a program which implements image compression on a computer.

REFERENCE SIGNS LIST 1 video encoding device 2 video encoding device 2A video encoding device 10 motion vector interpolation unit 11 detection unit 12 interpolation vector selection unit 13 transmission unit 14 reception unit 15 first candidate selection unit 16 second candidate selection unit 17 selection execution unit Reference Signs List 20 target selection unit 21 first target selection unit 22 second target selection unit 30 refinement unit 31 first refinement unit 32 second refinement unit 40 bi-prediction generation unit 50 prediction selection unit 60 compensation candidate generation unit 70 mode determination unit 80 vector storage Unit 81 first vector storage unit 82 second vector storage unit 90 motion vector compensation unit 91 detection unit 92 compensation vector selection unit 93 transmission unit 94 reception unit 95 first candidate selection unit 96 second candidate selection unit 97 selection execution unit 98 3 candidate selection unit 99 propagation condition determination unit 100 motion compensation unit 101 block division unit 102 pre-analysis unit 103 Intra prediction unit 104 Switching unit 105 Integer conversion unit 106 Quantization unit 107 Dequantization unit 108 Inverse integer conversion unit 109 Encoding unit 116 Filter unit 117 Frame storage unit 118 Subtraction unit 119 Addition unit 200 Motion compensation unit 1000 Computer 1001 Processor 1002 Memory 1003 Storage device 1004 I / O interface 1005 Recording medium 1010 Motion vector compensation circuit 1011 Detection circuit 1012 Compensation vector selection circuit 1013 Transmission circuit 1014 Receiving circuit 1015 First candidate selection circuit 1016 Second candidate selection circuit 1017 Selection execution circuit 1020 Target Selection circuit 1021 First object selection circuit 1022 Second object selection circuit 1030 Refine circuit 1031 First refine circuit 1032 Second refine circuit 1040 Measurement generation circuit 1050 prediction selection circuit 1060 compensation candidate generation circuit 1070 mode determination circuit 1080 vector storage circuit 1081 first vector storage circuit 1082 second vector storage circuit 1090 motion vector compensation circuit 1091 detection circuit 1092 compensation vector selection circuit 1093 transmission circuit 1094 reception Circuit 1095 First candidate selection circuit 1096 Second candidate selection circuit 1097 Selection execution circuit 1098 Third candidate selection circuit 1099 Propagation condition determination circuit 1100 Motion compensation circuit 1101 Block division circuit 1102 Pre-analysis circuit 1103 Intra prediction circuit 1104 Switching circuit 1105 Integer conversion Circuit 1106 Quantization circuit 1107 Inverse quantization circuit 1108 Inverse integer conversion circuit 1109 Encoding circuit 1116 Filter circuit 1117 Frame storage circuit 11 8 subtraction circuit 1119 adder circuit 1200 motion compensation circuit

Claims (10)

Among the pre-refinement vectors selected as a reference for searching for a predicted motion vector representing the prediction of the motion of the block into which the picture included in the video to be encoded is divided, the pre-refinement vector of the block adjacent to the current block. Supplementation candidate generating means for generating a candidate vector, comprising:
A motion vector compensating unit number of the refined before inheritance motion vectors obtained as a vector for the encoding target block is an accessory Hama vector Ru is compensated as the refining before vector when it is 1, selected from the candidate vectors,
A video encoding device comprising: The video encoding device according to claim 1, wherein the motion vector compensation unit selects the compensation vector based on a distance between the inherited motion vector and the candidate vector. The video encoding device according to claim 1, wherein the motion vector compensation unit selects the compensation vector based on a size of the candidate vector. 4. The motion vector compensation unit according to claim 1, wherein the motion vector compensation unit selects a plurality of secondary candidate vectors according to different conditions from the candidate vectors, and selects the compensation vector from the selected plurality of secondary candidate vectors. 5. A video encoding device according to claim 1. Among the pre-refinement vectors selected as a reference for searching for a predicted motion vector representing the prediction of the motion of the block into which the picture included in the video to be encoded is divided, the pre-refinement vector of the block adjacent to the current block. Generate a candidate vector containing
The refining front complement number of inheritance motion vectors obtained as a vector Ru is compensated as the refining prior vector in the case of 1 Hama vector of the encoding target block is selected from the candidate vectors,
Video encoding method. The video encoding method according to claim 5, wherein the compensation vector is selected based on a distance between the inherited motion vector and the candidate vector. The video encoding method according to claim 5, wherein the compensation vector is selected based on a size of the candidate vector. On the computer,
Among the pre-refinement vectors selected as a reference for searching for a predicted motion vector representing the prediction of the motion of the block into which the picture included in the video to be encoded is divided, the pre-refinement vector of the block adjacent to the current block. A compensation candidate generation process for generating a candidate vector, including:
A motion vector compensating process that the number of the refining before inheritance motion vectors obtained as a vector for the encoding target block is an accessory Hama vector Ru is compensated as the refining before vector when it is 1, selected from the candidate vectors,
A video encoding program that executes The video encoding program according to claim 8, wherein the motion vector compensation processing selects the compensation vector based on a distance between the inherited motion vector and the candidate vector. The video encoding program according to claim 8, wherein the motion vector compensation processing selects the compensation vector based on a size of the candidate vector.

Images