JP6677243B2 - Video encoding device, video encoding method and program - Google Patents

Description

  The present invention relates to a video encoding device, and more particularly to a circuit design technique in a video encoding method based on motion vector merging such as HEVC.

  Non-Patent Document 1 describes HEVC (High Efficiency Video Coding), which is a video coding scheme based on the H.265 standard recommended by the International Telecommunication Union Telecommunication Standardization Sector (ITU-T).

  In HEVC, each frame of a digitized video is divided into coding tree units (CTUs), and each CTU is coded in raster scan order. Each CTU is divided into coding units (CU: Coding Unit) in a quad tree structure and is coded. Each CU is divided into prediction units (PU: Prediction Unit) and predicted. Further, the prediction error of each CU is divided into transform units (TU: Transform Unit) in a quad tree structure and frequency-transformed. The largest CU is called the largest CU (LCU: Largest Coding Unit), and the smallest CU is called the smallest CU (SCU: Smallest Coding Unit).

  The CU is predictively encoded by intra prediction or inter-frame prediction (inter prediction).

  FIGS. 12A and 12B are explanatory diagrams showing an example of CU division when the CTU size is 64 × 64 (64 pixels × 64 pixels). FIG. 12A shows an example of a divided shape (hereinafter, also referred to as a block structure). FIG. 12B shows an example of a CU quad tree structure corresponding to the divided shape shown in FIG. 12A.

  The CU is divided into TUs in a quad tree structure. The way of division is the same as the case of CU division shown in FIG. 12A.

  FIG. 13 is an explanatory diagram showing a PU partitioning method in intra prediction and a PU partitioning method in inter prediction. FIG. 13 shows an example of a CU quad-tree structure of a CTU and candidates for PU division for each prediction mode.

  As shown in FIG. 13, when encoding is performed by intra prediction, a method of making the PU the same size as the CU size (2Nx2N) or dividing the PU into four sizes smaller than the CU size by one step (NxN) There are two division methods. However, the method of dividing into NxN is used only when the CU size is the minimum.

  As shown in FIG. 13, when encoding is performed by inter prediction, there is a method of setting the PU to the same size (2N × 2N) as the CU size. Alternatively, the CU is divided into two vertically symmetrical or left and right symmetrical rectangles (2NxN, Nx2N), and the CU is divided into two vertically asymmetrical or left and right asymmetrical rectangles (2NxnU, 2NxnD, nRx2N, nLx2N). There are four ways. That is, there are a total of seven division methods.

  In the case of inter prediction, coding based on motion compensation prediction is performed, and a motion vector is transmitted. The motion vector is transmitted for each PU. Therefore, the number of motion vectors per CTU depends on the CU quadtree structure. The number of motion vectors increases as the division becomes finer, and the code amount of the motion vectors increases.

  In the case where encoding is performed by intra prediction, when division is performed, the TU is sequentially divided starting from a PU that is a block of the same size as the CU or a PU that is a block into which the CU is divided into four. When encoding is performed by inter prediction, when division is performed, the TU is sequentially divided starting from the CU.

  As a method of transmitting a motion vector for each PU in inter prediction, there are two methods of adaptive motion vector prediction coding and merge coding. In the adaptive motion vector predictive coding, only a difference between a motion vector to be transmitted and a predicted motion vector obtained from a peripheral PU is transmitted. In the merge coding, only the indexes of the merge candidate list constructed by the motion vectors of the peripheral PUs are transmitted. A video encoding apparatus to which an encoding control technique related to merge index determination is applied is described in Patent Document 1.

  With reference to FIG. 14, the configuration and operation of a general video encoding device that outputs a bit stream using each CU of each frame of a digitized video as an input image will be described.

  FIG. 14 is a block diagram illustrating an example of a general video encoding device. The video encoding device 100 illustrated in FIG. 14 includes a transform unit 121, a quantization unit 122, an entropy encoding unit 127, an inverse quantization unit 123, an inverse transform unit 124, a buffer 125, a prediction unit 126, and an encoding parameter search unit. 110 is provided.

  The coding parameter search unit 110 calculates the coding costs for the CU quad tree structure / PU division shape / TU quad tree structure of the CTU, the prediction mode of the CU, the intra prediction direction of the intra PU, and the motion vector of the inter PU. calculate. The coding parameter search unit 110 searches for a motion vector of the inter PU by comparing the calculated coding costs, and determines a motion vector to be transmitted. The encoding parameter search unit 110 includes a motion vector search unit 111 that searches for a motion vector of an inter PU.

  The coding cost reflects a value related to the code amount and coding distortion (correlated with image quality). The encoding parameter search unit 110 uses the following RD (Rate Distortion) cost (Cost) as an example.

  Cost = D + λ · R

  D is a coding distortion, R is a code amount including the transform coefficient, and λ is a Lagrange multiplier.

  The encoding parameter search unit 110 determines, for each CTU, a CU quad tree structure / PU division shape / TU quad tree structure so as to increase the encoding efficiency according to the characteristics of the image.

  FIG. 15 is a block diagram illustrating an example of a general motion vector search unit. As illustrated in FIG. 15, the motion vector search unit 111 includes a motion vector search candidate generation unit 112, a merge candidate list output unit 113, and a motion vector evaluation unit 114.

  The motion vector evaluator 114 calculates an encoding cost for each of the motion vector output from the motion vector search candidate generator 112 and the motion vector output from the merge candidate list output unit 113. The motion vector evaluation unit 114 compares and evaluates the calculated coding costs, and selects a motion vector suitable for the target PU. The motion vector evaluation unit 114 outputs the selected motion vector.

  The prediction unit 126 generates a prediction signal for the CU input image signal based on the CU quad tree structure and the PU division shape determined by the coding parameter search unit 110. The prediction signal is generated based on intra prediction or inter prediction.

  The conversion unit 121 performs frequency conversion on a prediction error image (prediction error signal) obtained by subtracting a prediction signal from an input image signal based on the TU quadtree structure determined by the encoding parameter search unit 110. The transform unit 121 uses orthogonal transform of 4 × 4, 8 × 8, 16 × 16, or 32 × 32 block size based on frequency transform in transform coding of a prediction error signal. Specifically, DST (Discrete Sine Transform: Discrete Sine Transform) (integer precision) approximated by an integer operation is used for 4 × 4 TU of a luminance component of a CU to be intra-coded or inter-coded. For other TUs, a DCT (Discrete Cosine Transform) that is approximated by integer arithmetic (with integer precision) and that corresponds to the block size is used.

Quantization unit 122, an input transform coefficient (orthogonal transform coefficient) c _ij supplied from the conversion unit 121 and the quantization parameter Q _p, obtaining a quantization coefficient q _ij performs quantization process. q _ij is obtained by the following calculation.

q _ij = Int (c _ij / Q _step )
However, here
Q _step = (m _ij * 2 ^ q _bit ) / (Q _scale (Q _p % 6))
q _bit = 25 + (Q _p / 6) − BitDepth − log ₂ (N)
It is. Note that “%” represents a remainder operator.

Here, m _ij is a quantization weighting coefficient, Q _scale is a quantization step coefficient, BitDepth is the pixel bit accuracy of the input image, and N is the size of the orthogonal transform. Q _p larger the Q _step is increased, the code amount of the value q _ij resulting decreases.

  The inverse quantization unit 123 inversely quantizes the quantization coefficient. Further, the inverse transform unit 124 inversely transforms a result of the inverse quantization performed by the inverse quantization unit 123. The prediction error image obtained by the inverse transformation is supplied with a prediction signal and supplied to the buffer 125. The buffer 125 stores the supplied image as a reference image.

The code amount control unit (not shown) controls the encoding process so that the code amount as a result of encoding the frame being encoded becomes the target code amount. For example, the code amount control unit, by changing the quantization parameter Q _p, controls the code amount of the quantized coefficients. Further, the code amount control unit, the Lagrange multiplier λ by the function Q _p, can control block structure determination unit (not shown) through a Q _p.

  The merge candidate list output unit 113 outputs a list (merge candidate list) that can be used as a merge candidate in accordance with the procedure described in Section 8.5.3.2.1 of Non-Patent Document 1. The merge candidate list is a list constructed using a motion vector (hereinafter, referred to as a peripheral vector) of a spatially or temporally adjacent block.

  With reference to FIG. 16, the acquisition position of the peripheral vector will be described. FIG. 16 is an explanatory diagram illustrating the acquisition positions of the surrounding vectors of the encoding target prediction block. FIG. 16 shows the names of the respective positions of the peripheral blocks referred to when the merge candidate list is constructed.

  A rectangle shown at the center of FIG. 16 indicates a coding target prediction block. In addition, the circle illustrated in FIG. 16 indicates the acquisition position of the prediction block. There are seven positions A0, A1, B0, B1, B2, C0, and C1 at the acquisition positions of the peripheral prediction blocks shown in FIG. The peripheral vector is obtained from each of the seven acquisition positions.

  The procedure described in Non-Patent Document 1 specifies that up to five of the obtained peripheral vectors are listed as elements of a merge candidate list. The merge candidate list is generated by the merge candidate list output unit 113 according to the procedure described in Non-Patent Document 1. The number of elements in the merge candidate list is determined using the value MaxNumMergeCand given as an input to the encoding unit 120.

  FIG. 17 is a block diagram showing an outline of a general merge candidate list output unit. The merge candidate list output unit 113 illustrated in FIG. 17 includes a merge candidate list generation unit 200, a numCurrMergeCand storage unit 201, and an equivalence determination unit 202.

  The merge candidate list output unit 113 has a function of receiving MaxNumMergeCand as an input and outputting a candidate generation completion signal and a merge candidate list.

  The merge candidate list generation unit 200 has a function of generating a merge candidate list by adding the generated merge candidates to the merge candidate list. The merge candidate list generation unit 200 updates numCurrMergeCand stored in the numCurrMergeCand storage unit 201 while internally generating a merge candidate list.

  The numCurrMergeCand storage unit 201 has a function of storing numCurrMergeCand indicating the number of merge candidates included in the merge candidate list being generated.

  The equivalence determination unit 202 has a function of determining whether numCurrMergeCand stored in the numCurrMergeCand storage unit 201 is equal to MaxNumMergeCand. When numCurrMergeCand is equal to MaxNumMergeCand, the equivalence determination unit 202 outputs a candidate generation completion signal.

  Hereinafter, the operation when the merge candidate list output unit 113 shown in FIG. 17 generates the merge candidate list will be described with reference to FIG. FIG. 18 is a flowchart illustrating the operation of a process of generating a merge candidate list by a general merge candidate list output unit.

  The merge candidate list generation unit 200 generates one or more merge candidates and adds them to the merge candidate list (step S001). After the addition, the merge candidate list generation unit 200 updates numCurrMergeCand indicating the number of elements of the merge candidate list stored in the numCurrMergeCand storage unit 201 (step S002).

  The equivalence determination unit 202 confirms whether or not the updated numCurrMergeCand is equal to the input MaxNumMergeCand (step S003).

  If numCurrMergeCand and MaxNumMergeCand are equal (Yes in step S003), the merge candidate list output unit 113 ends the merge candidate list generation processing. If numCurrMergeCand and MaxNumMergeCand are not equal (No in step S003), the merge candidate list output unit 113 performs the processing in step S001 again.

  FIG. 19 is a block diagram illustrating an example of a general merge candidate list output unit 113. As shown in FIG. 19, the merge candidate list generation unit 200 includes a surrounding vector burst generation unit 210, a combined bi-prediction candidate generation unit 220, a zero merge candidate generation unit 230, a merge candidate list storage unit 240, and an increment unit 300. And an adder 301 and a subtractor 302.

  The peripheral vector burst generation unit 210 has a function of adding up to five generated peripheral vectors to the merge candidate list at a time. After the addition, the peripheral vector burst generation unit 210 updates numCurrMergeCand.

  The joint bi-prediction candidate generation unit 220 has a function of generating a joint bi-prediction candidate. The joint bi-prediction candidate is a motion vector newly generated by combining the neighboring vectors added to the merge candidate list. In the generation of combined bi-prediction candidates, the number of generated vectors depends on numCurrMergeCand. The combined bi-prediction candidate generation unit 220 generates a maximum of 12 vectors.

  The combined bi-prediction candidate generation unit 220 adds the combined bi-prediction candidate to the merge candidate list each time one combined bi-prediction candidate is generated. The combined bi-prediction candidate generation unit 220 updates numCurrMergeCand by using the increment unit 300 each time one combined bi-prediction candidate is generated.

  The zero merge candidate generation unit 230 has a function of generating a zero merge candidate. The zero merge candidate generation unit 230 generates a zero merge candidate when numCurrMergeCand is less than MaxNumMergeCand. The zero merge candidate generation unit 230 generates zero merge candidates by the difference between them so that numCurrMergeCand becomes equal to MaxNumMergeCand, and adds the generated zero merge candidates to the merge candidate list.

  Hereinafter, an operation when the merge candidate list output unit 113 illustrated in FIG. 19 generates the merge candidate list will be described with reference to FIG. FIG. 20 is a flowchart showing a specific operation of the process of generating a merge candidate list by the general merge candidate list output unit 113. The operation illustrated in FIG. 20 is an operation in a case where the merge candidate list is generated according to the procedure defined in the above standard.

  The peripheral vector burst generation unit 210 enumerates a maximum of five peripheral vectors to be elements of the merge candidate list (step S011). The peripheral vector burst generation unit 210 adds the listed peripheral vectors to the merge candidate list at once according to the above procedure (step S012).

  At the time of addition, the peripheral vector burst generation unit 210 calculates the number of elements in the merge candidate list, and updates numCurrMergeCand stored in the numCurrMergeCand storage unit 201 (step S013). The equivalence determination unit 202 determines whether numCurrMergeCand is equal to MaxNumMergeCand (step S014).

  When numCurrMergeCand and MaxNumMergeCand are equal (Yes in step S014), the equivalence determination unit 202 outputs a candidate generation completion signal indicating that the generation of the merge candidate list has been completed. The merge candidate list output unit 113 ends the merge candidate list generation processing.

  When numCurrMergeCand is not equal to MaxNumMergeCand (No in step S014), for example, when numCurrMergeCand is less than MaxNumMergeCand, the merge candidate list generation unit 200 performs a process of generating a combined bi-prediction candidate and an additional process.

  The combined bi-prediction candidate generation unit 220 newly generates one vector that is a combined bi-prediction candidate (step S015). Each time one combined vector is generated, the combined bi-prediction candidate generating unit 220 adds the generated vector to the merge candidate list (step S016).

  Next, the combined bi-prediction candidate generation unit 220 updates numCurrMergeCand (Step S017). After numCurrMergeCand is updated, the equivalence determination unit 202 determines whether numCurrMergeCand is equal to MaxNumMergeCand (step S018).

  If numCurrMergeCand is equal to MaxNumMergeCand (Yes in step S018), the equivalence determination unit 202 outputs a candidate generation completion signal indicating that the generation of the merge candidate list has been completed. The merge candidate list output unit 113 ends the merge candidate list generation processing.

  If numCurrMergeCand and MaxNumMergeCand are not equal (No in step S018), the combined bi-prediction candidate generation unit 220 checks whether the generation processing and the addition processing of the combined bi-prediction candidate have been completed (step S019).

  For example, when the number of combined bi-prediction candidates generated reaches a predetermined value, the combined bi-prediction candidate generation unit 220 ends the combined bi-prediction candidate generation processing and the addition processing. If the generation processing and the addition processing of the combined bi-prediction candidate have not been completed (No in step S019), the combined bi-prediction candidate generation unit 220 performs the processing of step S015 again.

  If the generation processing and the addition processing of the combined bi-prediction candidate have been completed (Yes in step S019), numCurrMergeCand is still smaller than MaxNumMergeCand. Therefore, the zero merge candidate generation unit 230 performs a generation process and an addition process of the zero merge candidate.

  In the generation process of the zero merge candidate, the zero merge candidate generation unit 230 generates the zero merge candidate by the difference between the number of elements in the merge candidate list and MaxNumMergeCand (Step S020). The zero merge candidate generation unit 230 adds the generated zero merge candidate to the end of the merge candidate list (Step S021).

  After the zero merge candidate is added, the equivalence determination unit 202 outputs a candidate generation completion signal indicating that the generation of the merge candidate list has been completed. The merge candidate list output unit 113 ends the merge candidate list generation processing.

  Upon receiving the notification of the completion of the generation of the merge candidate list, the motion vector evaluation unit 114 starts processing using the generated merge candidate list.

  Patent Document 2 discloses a technique for determining a set of motion vector prediction candidates in a motion vector prediction process.

Japanese Patent No. 5590269 JP 2014-517658 A

ITU-T recommendation H.265 High efficiency video coding, April 2013

  According to the merge candidate list generation procedure shown in FIG. 20, the time required for generating the merge candidate list may vary. Further, the time required to generate the merge candidate list becomes longer. Further, the motion vector evaluation unit 114 cannot start the processing until the processing of generating the merge candidate list ends.

  FIG. 21 is an explanatory diagram showing the processing time of a general motion vector search unit 111. FIG. 21 shows the processing time of the motion vector search unit 111 when the merge candidate list output process and the motion vector evaluation process are not performed in parallel.

  As shown in FIG. 21, while the merge candidate list output unit 113 is generating the merge candidate list, the motion vector evaluation unit 114 cannot operate. The merge candidate list output unit 113 outputs a generation completion signal after completing generation of the merge candidate list. When a generation completion signal is input, the motion vector evaluation unit 114 can start evaluation processing on the merge candidate list.

  When the time of the merge candidate list generation processing by the merge candidate list output unit 113 increases, the time during which the motion vector evaluation unit 114 cannot operate also increases. As a result, the processing time of the entire motion vector search unit 111 becomes longer. As described above, the general video encoding device illustrated in FIG. 14 has a problem that the generation time of the merge candidate list is long and the time until the start of the process using the merge candidate list is long.

  Therefore, an object of the present invention is to provide a video encoding device, a video encoding method, and a program recording medium that can accelerate the start of processing using a merge candidate list.

  A video encoding device according to the present invention includes a merging candidate list generating unit configured to generate a merging candidate list from an adjacent block, and a process that uses the merging candidate list when the merging candidate list includes at least one merging candidate. Output means for outputting a process start signal to be started.

  In the video encoding method according to the present invention, a merge candidate list is generated from adjacent blocks, and when at least one merge candidate is included in the merge candidate list, a process start signal for starting a process using the merge candidate list is generated. It is characterized by outputting.

  In the program recording medium according to the present invention, a generation process for generating a merge candidate list from adjacent blocks, and a process for starting a process in which the merge candidate list is used when at least one merge candidate is included in the merge candidate list A program for causing a computer to execute an output process of outputting a signal is recorded.

  According to the present invention, it is possible to hasten the start of the process in which the merge candidate list is used.

FIG. 1 is a block diagram showing an outline of a merge candidate list output unit according to the present invention. FIG. 2 is a flowchart showing the operation of the merge candidate list generation processing by the merge candidate list output unit according to the present invention. FIG. 3 is an explanatory diagram showing the processing time of the motion vector search unit according to the present invention. FIG. 4 is a block diagram showing a configuration example of the first embodiment of the merge candidate list output unit according to the present invention. FIG. 5 is a flowchart illustrating a specific operation of the merge candidate list generation processing by the merge candidate list output unit according to the first embodiment. FIG. 6 is a block diagram showing a configuration example of the second embodiment of the merge candidate list output unit according to the present invention. FIG. 7 is a flowchart illustrating a specific operation of a merge candidate list generation process by the merge candidate list output unit according to the second embodiment. FIG. 8 is a block diagram showing a configuration example of the third embodiment of the merge candidate list output unit according to the present invention. FIG. 9 is a flowchart illustrating a specific operation of the merge candidate list generation processing by the merge candidate list output unit according to the third embodiment. FIG. 10 is a block diagram illustrating a configuration example of an information processing system capable of realizing the functions of the video encoding device according to the present invention. FIG. 11 is a block diagram showing an outline of a video encoding device according to the present invention. FIG. 12A is an explanatory diagram showing an example of the CU division shape when the CTU size is 64 × 64 (64 pixels × 64 pixels). FIG. 12B is an explanatory diagram showing an example of a CU quad tree structure when the CTU size is 64 × 64 (64 pixels × 64 pixels). FIG. 13 is an explanatory diagram showing a PU partitioning method in intra prediction and a PU partitioning method in inter prediction. FIG. 14 is a block diagram illustrating an example of a general video encoding device. FIG. 15 is a block diagram illustrating an example of a general motion vector search unit. FIG. 16 is an explanatory diagram illustrating the acquisition positions of the surrounding vectors of the encoding target prediction block. FIG. 17 is a block diagram showing an outline of a general merge candidate list output unit. FIG. 18 is a flowchart illustrating the operation of a process of generating a merge candidate list by a general merge candidate list output unit. FIG. 19 is a block diagram illustrating an example of a general merge candidate list output unit. FIG. 20 is a flowchart illustrating a specific operation of a process of generating a merge candidate list by a general merge candidate list output unit. FIG. 21 is an explanatory diagram showing the processing time of a general motion vector search unit.

[Description of configuration]
Hereinafter, the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an outline of the merge candidate list output unit 113 according to the present invention. Note that arrows shown in the block diagrams of FIG. 1 and subsequent drawings show an example of a moving direction of information. The moving direction of the information is not limited to the illustrated direction.

  The configuration of the merge candidate list output unit 113 shown in FIG. 1 is different from the configuration of the merge candidate list output unit 113 shown in FIG. 17 in that a non-zero determination unit 203 is added. The configuration of the merge candidate list output unit 113 shown in FIG. 1 other than the non-zero determination unit 203 is the same as the configuration of the merge candidate list output unit 113 shown in FIG.

  The non-zero determination unit 203 has a function of determining whether numCurrMergeCand stored in the numCurrMergeCand storage unit 201 is not 0. When determining that numCurrMergeCand is not 0, the non-zero determination unit 203 outputs an evaluation start enable signal.

[Description of operation]
Hereinafter, the operation when the merge candidate list output unit 113 shown in FIG. 1 generates the merge candidate list will be described with reference to FIG. FIG. 2 is a flowchart showing the operation of the merge candidate list generation processing by the merge candidate list output unit 113 according to the present invention.

  The processing in steps S101 to S102 is the same as the processing in steps S001 to S002 shown in FIG.

  After numCurrMergeCand is updated, the non-zero determination unit 203 determines whether numCurrMergeCand is not 0 (step S103).

  If numCurrMergeCand is not 0 (Yes in step S103), the non-zero determination unit 203 outputs an evaluation start enable signal (step S104).

  After the processing in step S103 or S104, the equivalence determination unit 202 checks whether numCurrMergeCand and MaxNumMergeCand stored in the numCurrMergeCand storage unit 201 are equal (step S105).

  The process in step S105 is the same as the process in step S003 shown in FIG.

[Explanation of effects]
Hereinafter, the effects of the present invention will be described. According to the present invention, before the generation of the merge candidate list is completed, the non-zero determination unit 203 outputs an evaluation start enable signal. Therefore, the motion vector evaluation unit 114 that performs the subsequent processing can start the processing without waiting for the completion of the generation of the merge candidate list by receiving the evaluation start enable signal. That is, the merge candidate list output unit 113 can advance the processing start time of the motion vector evaluation unit 114 that performs processing using the merge candidate list.

  FIG. 3 is an explanatory diagram showing the processing time of the motion vector search unit according to the present invention. FIG. 3 shows the processing time of the motion vector search unit when the merge candidate list output process and the motion vector evaluation process are performed in parallel.

  As shown in FIG. 21, in the normal motion vector search unit, the motion vector evaluation processing is started when the generation of the merge candidate list is completed. However, as shown in FIG. 3, the merge candidate list output unit 113 in the present embodiment notifies that the processing can be started when at least one merge candidate is added to the merge candidate list. Therefore, the motion vector evaluation unit 114 can operate in parallel with the merge candidate list output unit 113. That is, since the motion vector evaluation process is started during the generation of the merge candidate list, the processing time of the entire motion vector search unit is reduced.

  Hereinafter, a specific configuration and operation of the merge candidate list output unit 113 in each embodiment of the present invention will be described.

Embodiment 1 FIG.
[Description of configuration]
Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 4 is a block diagram showing a configuration example of the first embodiment of the merge candidate list output unit 113 according to the present invention.

  The configuration of the merge candidate list output unit 113 illustrated in FIG. 4 is different from the configuration of the merge candidate list output unit 113 illustrated in FIG. 19 in that a non-zero determination unit 203 is added. The configuration of the merge candidate list output unit 113 shown in FIG. 4 other than the non-zero determination unit 203 is the same as the configuration of the merge candidate list output unit 113 shown in FIG.

[Description of operation]
Hereinafter, the operation of the merge candidate list output unit 113 according to the present embodiment will be described with reference to FIG. FIG. 5 is a flowchart illustrating a specific operation of the merge candidate list generation processing by the merge candidate list output unit 113 according to the first embodiment.

When the generation of the merge candidate list is started, the peripheral vector burst generation unit 210 generates a maximum of five peripheral vectors at a time (step S111). The peripheral vector burst generation unit 210 adds the generated peripheral vectors to the merge candidate list at once (step S112).
After the addition, the peripheral vector burst generation unit 210 updates numCurrMergeCand (step S113).

  After numCurrMergeCand is updated, the non-zero determination unit 203 determines whether numCurrMergeCand is not 0 (step S114).

  If numCurrMergeCand is not 0 (Yes in step S114), the non-zero determination unit 203 outputs an evaluation start enable signal (step S115).

  After the processing in step S114 or S115, the equivalence determination unit 202 checks whether numCurrMergeCand and MaxNumMergeCand stored in the numCurrMergeCand storage unit 201 are equal (step S116).

  When numCurrMergeCand is equal to MaxNumMergeCand (Yes in step S116), the equivalence determination unit 202 outputs a candidate generation completion signal indicating that the generation of the merge candidate list has been completed. The merge candidate list output unit 113 ends the merge candidate list generation processing.

  If numCurrMergeCand and MaxNumMergeCand are not equal (No in step S116), the combined bi-prediction candidate generation unit 220 generates a combined bi-prediction candidate (step S117). Each time one combined bi-prediction candidate is generated, the combined bi-prediction candidate generation unit 220 adds the generated combined bi-prediction candidate to the merge candidate list (step S118). Next, the combined bi-prediction candidate generation unit 220 updates numCurrMergeCand (step S119).

  Each time one combined bi-prediction candidate is added to the merge candidate list and numCurrMergeCand is updated, the non-zero determining unit 203 determines whether numCurrMergeCand is not 0 (step S120). If numCurrMergeCand is not 0 (Yes in step S120), the non-zero determination unit 203 outputs an evaluation start enable signal (step S121).

  After the processing of step S120 or S121, the equivalence determination unit 202 checks whether numCurrMergeCand and MaxNumMergeCand stored in the numCurrMergeCand storage unit 201 are equal (step S122).

  When numCurrMergeCand is equal to MaxNumMergeCand (Yes in step S122), the equivalence determination unit 202 outputs a candidate generation completion signal indicating that the generation of the merge candidate list has been completed. The merge candidate list output unit 113 ends the merge candidate list generation processing.

  When numCurrMergeCand and MaxNumMergeCand are not equal (No in step S122), the combined bi-prediction candidate generation unit 220 checks whether the generation processing and the addition processing of the combined bi-prediction candidate have been completed (step S123).

  When the generation processing and the addition processing of the combined bi-prediction candidate have not been completed (No in step S123), the combined bi-prediction candidate generation unit 220 performs the processing of step S117 again.

  When the generation processing and the addition processing of the combined bi-prediction candidate have been completed (Yes in step S123), the zero merge candidate generation unit 230 generates a zero vector corresponding to the zero merge candidate (step S124).

  The zero merge candidate generation unit 230 creates a zero vector by the difference between numCurrMergeCand and MaxNumMergeCand such that numCurrMergeCand is equal to MaxNumMergeCand. The zero merge candidate generation unit 230 adds the created zero vector to the merge candidate list (Step S125).

  When the zero vector addition process is completed, the equivalence determination unit 202 outputs a candidate generation completion signal indicating that the generation of the merge candidate list has been completed. The merge candidate list output unit 113 ends the merge candidate list generation processing.

[Explanation of effects]
A video encoding device that performs motion compensation prediction encoding based on a merge candidate list according to the present embodiment includes a merge candidate list generation unit and a non-zero determination unit. During the generation of the merge candidate list, the merge candidate list generation unit updates the number of elements in the merge candidate list. When the number of elements in the updated merge candidate list is not 0, the non-zero determination unit outputs a signal indicating that subsequent processing can be started. As a result, the time until the start of the process using the merge candidate list is reduced.

  The merge candidate list output unit 113 according to the present embodiment outputs a merge candidate list including peripheral vectors in a video encoding device having an encoding unit that performs encoding using merge candidates. The merge candidate list output unit 113 updates numCurrMergeCand each time a candidate is added, and outputs a process start enable signal to the motion vector evaluation unit 114 that performs the subsequent process when numCurrMergeCand is not 0. The merge candidate list output unit 113 in this embodiment can hasten the start of the subsequent processing by the motion vector evaluation unit 114.

  The motion vector search unit 111 in the present embodiment can execute the merge candidate list generation process by the merge candidate list output unit 113 and the evaluation process by the motion vector evaluation unit 114 in parallel. That is, the motion vector search unit 111 can reduce the processing time required for the entire process from the start of the merge candidate list generation process to the completion of the motion vector evaluation process.

Embodiment 2. FIG.
[Description of configuration]
Next, a second embodiment of the present invention will be described with reference to the drawings. FIG. 6 is a block diagram showing a configuration example of the second embodiment of the merge candidate list output unit 113 according to the present invention.

  The configuration of the merge candidate list output unit 113 shown in FIG. 6 is different from the configuration of the merge candidate list output unit 113 shown in FIG. 4 in that the peripheral vector burst generation unit 210 is replaced by a peripheral vector sequential generation unit 410. Are different.

  Also, the difference is that the entity that updates numCurrMergeCand when a peripheral vector is generated has been changed from the peripheral vector burst generation unit 210 to an increment unit 300 that adds one. The configuration of the merge candidate list output unit 113 shown in FIG. 6 other than the peripheral vector sequential generation unit 410 and the increment unit 300 is the same as the configuration of the merge candidate list output unit 113 shown in FIG.

  The peripheral vector successive generation unit 410 adds the generated peripheral vector to the merge candidate list every time one peripheral vector is generated. Further, the increment unit 300 updates numCurrMergeCand every time one peripheral vector is added to the merge candidate list.

[Description of operation]
Hereinafter, the operation of the merge candidate list output unit 113 of the present embodiment will be described with reference to FIG. FIG. 7 is a flowchart illustrating a specific operation of the merge candidate list generation processing by the merge candidate list output unit 113 according to the second embodiment.

  When the generation of the merge candidate list is started, the peripheral vector sequential generation unit 410 generates one peripheral vector (step S211). The peripheral vector successive generation unit 410 adds one generated peripheral vector to the merge candidate list (step S212). After the addition, the increment unit 300 updates numCurrMergeCand (step S213).

  After numCurrMergeCand is updated, the non-zero determining unit 203 determines whether numCurrMergeCand is not 0 (step S214).

  If numCurrMergeCand is not 0 (Yes in step S214), the non-zero determination unit 203 outputs an evaluation start enable signal (step S215).

  After the processing in step S214 or S215, the peripheral vector sequential generation unit 410 checks whether or not the peripheral vector generation processing has been completed according to the procedure defined in the standard (step S216). If the peripheral vector generation processing has not been completed (No in step S216), the peripheral vector sequential generation unit 410 performs the processing in step S211 again.

  If the generation processing of the peripheral vector has been completed (Yes in step S216), the equivalence determination unit 202 checks whether numCurrMergeCand stored in the numCurrMergeCand storage unit 201 is equal to MaxNumMergeCand (step S217).

  The processing in steps S217 to S226 is the same as the processing in steps S116 to S125 shown in FIG.

[Explanation of effects]
Hereinafter, effects of the present embodiment will be described. According to the merge candidate list output unit 113 of the present embodiment, the non-zero determination unit 203 outputs an evaluation start enable signal during the generation and addition processing of the peripheral vector. Therefore, the motion vector evaluation unit 114 that performs the subsequent processing can start the processing earlier than in the first embodiment. That is, in the motion vector search unit 111 of the present embodiment, the degree of parallelism between the merge candidate list generation process by the merge candidate list output unit 113 and the evaluation process by the motion vector evaluation unit 114 may be higher than in the first embodiment. There is.

Embodiment 3 FIG.
[Description of configuration]
Next, a third embodiment of the present invention will be described with reference to the drawings. FIG. 8 is a block diagram showing a configuration example of the third embodiment of the merge candidate list output unit 113 according to the present invention.

  The configuration of the merge candidate list output unit 113 shown in FIG. 8 is different from the configuration of the merge candidate list output unit 113 shown in FIG. 6 in that the zero merge candidate generation unit 230 is replaced by an initialization zero merge candidate generation unit 430. Are different.

  The configuration of the merge candidate list output unit 113 shown in FIG. 8 differs from the configuration shown in FIG. 6 in that a zero merge candidate shift unit 440 is added. The configurations of the numCurrMergeCand storage unit 201, the combined bi-prediction candidate generation unit 220, the merge candidate list storage unit 240, the increment unit 300, and the peripheral vector sequential generation unit 410 illustrated in FIG. It is the same as that of the component.

  The initialization zero merge candidate generation unit 430 operates immediately after the operation of the merge candidate list generation unit 200 starts. The initialization zero merge candidate generation unit 430 initializes the merge candidate list by adding a zero merge candidate defined in the standard to the merge candidate list.

  Each time the peripheral vector successive generation unit 410 and the combined bi-prediction candidate generation unit 220 add a candidate to the merge candidate list, the zero merge candidate shift unit 440 converts the zero merge candidate added to the merge candidate list by initialization into a merge candidate. Shift to the end of the list.

[Description of operation]
Hereinafter, the operation of the merge candidate list output unit 113 of the present embodiment will be described with reference to FIG. FIG. 9 is a flowchart illustrating a specific operation of the merge candidate list generation processing by the merge candidate list output unit 113 according to the third embodiment.

  Prior to the start of the merge candidate list generation process, the initialized zero merge candidate generation unit 430 generates all the zero merge candidates specified in the standard (step S311). The initialized zero merge candidate generation unit 430 adds the generated zero merge candidate to the merge candidate list (Step S312).

  When the generated zero merge candidate is added to the merge candidate list, numCurrMergeCand is not updated. Even if a zero merge candidate is added to the merge candidate list, numCurrMergeCand remains at the initial value 0.

In FIG. 9, the processing from step S313 corresponds to the merge candidate list generation processing.
The processing in steps S313 to S326 other than step S315 and step S322 is the same as the processing in steps S211 to S224 shown in FIG.

  When numCurrMergeCand is updated after the merge candidate is added to the merge candidate list, the zero merge candidate shift unit 440 shifts the zero vector in the merge candidate list to the end (steps S315 and S322).

  After the generation process and the addition process of the combined bi-prediction candidate are completed (Yes in step S326), the equivalence determination unit 202 outputs a candidate generation completion signal indicating that the generation of the merge candidate list has been completed. The merge candidate list output unit 113 ends the merge candidate list generation processing.

[Explanation of effects]
Hereinafter, effects of the present embodiment will be described. The merge candidate list output unit 113 of the present embodiment performs the zero vector generation processing and the addition processing that were performed last in the first embodiment and the second embodiment before starting the merge candidate list generation processing. Do. As a result, the worst value of the time required for outputting the evaluation start enable signal is reduced. That is, when neither the surrounding vector nor the combined bi-prediction candidate exists, and only the zero vector is added to the merge candidate list, the time required for generating the merge candidate list is reduced.

  Further, each of the above embodiments can be configured by hardware, but can also be realized by, for example, a computer program recorded on a recording medium.

  The information processing system shown in FIG. 10 includes a processor 1001, a program memory 1002, a storage medium (recording medium) 1003 for storing video data, and a storage medium 1004 for storing data such as a bit stream. The storage medium 1003 and the storage medium 1004 may be separate storage media or different storage areas of the same storage medium. As the storage medium, a magnetic storage medium such as a hard disk can be used. In the storage medium 1003, at least the area where the program is stored is a non-transitory tangible storage area.

  In the information processing system shown in FIG. 10, the program memory 1002 stores a program for realizing the function of each block shown in FIGS. 1, 4, 6, and 8. The processor 1001 executes the processing according to the program stored in the program memory 1002, thereby realizing the function of the merge candidate list output unit 113 shown in each of FIGS. 1, 4, 6, and 8. I do.

  Next, the outline of the present invention will be described. FIG. 11 is a block diagram showing an outline of a video encoding device according to the present invention. The video encoding device 10 according to the present invention includes a merge candidate list generation unit 11 (for example, a merge candidate list generation unit 200) and an output unit 12 (for example, a non-zero determination unit 203). The merging candidate list generating means 11 generates a merging candidate list from adjacent blocks. When at least one merge candidate is included in the merge candidate list, the output unit 12 outputs a process start signal for starting a process using the merge candidate list.

  With such a configuration, the video encoding device 10 can hasten the start of the process in which the merge candidate list is used.

  In addition, the merge candidate list generation unit 11 generates a peripheral vector that is a merge candidate, and adds a peripheral vector to the merge candidate list every time one peripheral vector is generated (for example, a peripheral vector sequential generation unit). Unit 410).

  With such a configuration, the video encoding device can further increase the degree of parallelism between the merge candidate generation process and the motion vector evaluation process.

  Further, the merge candidate list generation unit 11 may include an initialization zero merge candidate generation unit (for example, an initialization zero merge candidate generation unit 430) and a zero merge candidate shift unit (for example, a zero merge candidate shift unit 440). The initialization zero merge candidate generation means may initialize the merge candidate list with the zero merge candidates before starting generation of the merge candidate list. The zero merge candidate shifting means may shift the zero merge candidate to the end of the merge candidate list when the merge candidate is added to the merge candidate list.

  With such a configuration, the video encoding device 10 can reduce the worst value of the time required to output the processing start signal.

  INDUSTRIAL APPLICABILITY The present invention is applicable to a video compression device and a program for realizing video compression by a computer.

  The present invention has been described above using the above-described embodiment as a typical example. However, the invention is not limited to the embodiments described above. That is, the present invention can apply various aspects that can be understood by those skilled in the art within the scope of the present invention.

  This application claims priority based on Japanese Patent Application No. 2005-055983 filed on March 19, 2015, and incorporates the entire disclosure thereof.

10, 100 Video encoding device 11 Merge candidate list generation means 12 Output means 110 Encoding parameter search unit 111 Motion vector search unit 112 Motion vector search candidate generation unit 113 Merge candidate list output unit 114 Motion vector evaluation unit 120 Encoding unit 121 Transformation unit 122 Quantization unit 123 Inverse quantization unit 124 Inverse transformation unit 125 Buffer 126 Prediction unit 127 Entropy encoding unit 200 Merge candidate list generation unit 201 numCurrMergeCand storage unit 202 Equivalence determination unit 203 Non-zero determination unit 210 Peripheral vector burst generation unit 220 Combined bi-prediction candidate generation unit 230 Zero merge candidate generation unit 240 Merge candidate list storage unit 300 Increment unit 301 Addition unit 302 Subtraction unit 410 Neighboring vector sequential generation unit 430 Initialization zero merge candidate generation unit 440 Zero merge candidate shift Unit 1001 processor 1002 program memory 1003, 1004 storage medium

Claims (9)

Merge candidate list generating means for generating a merge candidate list from adjacent blocks;
If merging candidate to the merging candidate list contains at least one, and output means for outputting the evaluation process in the vector evaluation unit motion processing start signal to start for the merging candidate list,
The video encoding apparatus according to claim 1, wherein said motion vector evaluator receives said processing start signal and executes said evaluation processing in parallel with generation of said merge candidate list by said merge candidate list generator . The merging candidate list generating unit includes a peripheral vector successively generating unit that generates a peripheral vector that is the merge candidate, and adds the peripheral vector to the merge candidate list each time one peripheral vector is generated. A video encoding device according to claim 1. The merge candidate list generation unit includes an initialization zero merge candidate generation unit and a zero merge candidate shift unit,
The initialization zero merge candidate generation means generates a zero merge candidate before starting generation of the merge candidate list, and initializes the merge candidate list with the generated zero merge candidate,
3. The video encoding device according to claim 1, wherein the zero merge candidate shift unit shifts the zero merge candidate to a tail side of the merge candidate list when the merge candidate is added to the merge candidate list. 4. Generate a merge candidate list from adjacent blocks,
If the merge candidate list M a di candidates contain at least one, and outputs a processing start signal for starting the evaluation process for the merging candidate list,
Receiving the output processing start signal, performing the evaluation processing in parallel with the generation of the merge candidate list . Generating a peripheral vector that is a candidate for the merge,
The video encoding method according to claim 4, wherein each time one of the peripheral vectors is generated, the peripheral vector is added to the merge candidate list. Initializing the merge candidate list with zero merge candidates before the generation of the merge candidate list starts,
The video encoding method according to claim 4 or 5, wherein when the merge candidate is added to the merge candidate list, the zero merge candidate is shifted to the end of the merge candidate list. On the computer,
A generation process of generating a merge candidate list from an adjacent block; and, when the merge candidate list includes at least one merge candidate, outputting a process start signal to start an evaluation process for the merge candidate list to the motion vector evaluation process. an output process of is executed,
The motion vector evaluation processing is a program that, upon receiving the processing start signal, executes the evaluation processing in parallel with the generation of the merge candidate list by the merge candidate list generation processing . On the computer,
The program according to claim 7, wherein a generation process of generating a peripheral vector that is the merge candidate and an additional process of adding the peripheral vector to the merge candidate list each time one of the peripheral vectors is generated are executed. On the computer,
An initialization process of generating a zero merge candidate before the generation start of the merge candidate list and initializing the merge candidate list with the generated zero merge candidate, and when the merge candidate is added to the merge candidate list, The non-transitory computer-readable storage medium according to claim 7, wherein a shift process of shifting a zero merge candidate to the end of the merge candidate list is performed.

Images