JP6323185B2 - Moving picture coding apparatus, moving picture coding method, and moving picture coding computer program - Google Patents

Description

  The present invention relates to, for example, a moving picture coding apparatus, a moving picture coding method, and a moving picture coding computer program for coding a picture with reference to another coded picture.

  The moving image data generally has a very large amount of data. For this reason, a device that handles moving image data encodes moving image data with high efficiency when moving image data is to be transmitted to another device or when moving image data is to be stored in a storage device. “High-efficiency encoding” refers to an encoding process for converting a data string into another data string and compressing the data amount.

  As a high-efficiency encoding method for moving image data, an intra-picture prediction (intra prediction) encoding method is known. This encoding method uses the fact that moving image data is highly correlated in the spatial direction, and does not use encoded images of other pictures. A picture encoded by the intra-picture predictive encoding method can be restored only with information in the picture.

  Further, as another encoding method employed in the high efficiency encoding method, an inter-picture prediction (inter prediction) encoding method is known. This encoding method uses the fact that moving image data is highly correlated in the time direction. In moving image data, in general, there is often a high degree of similarity between a picture at a certain timing and a picture that follows that picture. Therefore, the inter prediction encoding method uses the property of moving image data. In general, a moving image encoding apparatus divides an original picture to be encoded into a plurality of blocks. For each block, the moving image encoding apparatus executes a process called motion search process, and selects an area similar to that block as a reference area from a reference picture obtained by decoding an encoded picture. To do. Then, the moving image coding apparatus removes temporal redundancy by calculating a prediction error block representing a difference between the reference region and the block. Then, the moving image encoding apparatus realizes a high compression rate by encoding the motion vector information indicating the reference region and the prediction error block. In general, the inter prediction encoding method has higher compression efficiency than the intra prediction encoding method.

  In the motion search process, the moving image encoding apparatus obtains a motion vector by performing block matching with a reference picture and specifying a reference region for an encoding target block. In this block matching, since an area wider than the encoding target block on the reference picture is a search range in which block matching is performed, the memory access amount representing the amount of data read from the memory in which the reference picture is stored is large. Therefore, the ratio of the memory access amount by the motion search process to the memory access amount in the entire inter prediction encoding process is large.

  Further, it is assumed that the latest video coding scheme (High Efficiency Video Coding, HEVC) handles video data having a high resolution such as 4K UHD. When the moving image encoding apparatus encodes moving image data having such a high resolution, the number of blocks included in one picture increases, so that the memory access amount also increases. As the memory access amount increases, the power required for executing the moving image encoding process increases and the latency increases. Therefore, it is preferable that the memory access amount of the motion search process can be reduced.

  For example, Patent Document 1 proposes an image encoding device that generates an encoded image that can reduce the data transfer amount of a reference image during motion compensation processing in an image decoding device. This image encoding apparatus sets a search range that is a transfer amount with a small transfer amount of data generated in motion compensation processing at the time of decoding an encoded image, and acquires pixel data of the search range from the image storage unit. To predict motion.

International Publication No. 10/140338

  However, in the image encoding device disclosed in Patent Document 1, the search range is too narrowly limited, and within the search range, a reference region with a sufficiently small prediction error for the encoding target block cannot be set and motion search is performed. Accuracy may be reduced. In this case, the number of pixels having a value other than 0 in the prediction error block increases, and the number of orthogonal transform coefficients having a value other than 0 also exists in the set of orthogonal transform coefficients obtained by orthogonal transform of the prediction error block. Become more. Therefore, in order to keep the picture quality of a picture to be reproduced at a certain level or more, the quantization width that can be applied to the orthogonal transform coefficient is reduced, and as a result, the encoding efficiency is lowered.

  Therefore, an object of the present specification is to provide a moving image encoding apparatus that can reduce the memory access amount in the motion search process while suppressing a decrease in encoding efficiency.

  According to one embodiment, a moving image code that encodes an error between an encoding target block on a picture included in moving image data and a reference block indicated by a motion vector on another encoded picture A device is provided. This moving image encoding device is defined by the search mode on the reduced picture obtained by reducing the number of pixels included in the other encoded pictures for each of the plurality of search modes that define the search method of the motion vector. In the region corresponding to the search range of the motion vector to be obtained, the region that most closely matches the reduced block obtained by reducing the number of pixels included in the encoding target block is obtained, and the search range is calculated from the amount of movement between the reduced block and the most coincident region. A motion vector candidate that identifies the coding cost, a coding cost calculation unit that calculates a coding cost that represents the coding amount of the coding target block, and a motion vector for each of a plurality of search modes based on a search range in the search mode To calculate the memory access amount indicating the amount of data read from the storage unit storing other pictures. An amount calculation unit, a search mode determination unit that selects a search mode to be applied to the encoding target block from the plurality of search modes based on the memory access amount and the encoding cost for each of the plurality of search modes, and a selection A search range is set on another encoded picture according to the search mode specified, and pixel data included in the search range on the other encoded picture is read from the storage unit, and encoded within the search range. A motion vector calculating unit that obtains a region that most closely matches the encoding target block and calculates a vector representing a movement amount between the encoding target block and the most matching region as a motion vector;

The objects and advantages of the invention will be realized and attained by means of the elements and combinations particularly pointed out in the appended claims.
It should be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are not restrictive of the invention as claimed.

  The moving picture encoding apparatus disclosed in this specification can reduce the memory access amount in the motion search process while suppressing a decrease in encoding efficiency.

It is a figure which shows an example of the division | segmentation of the picture by HEVC. It is a schematic block diagram of the moving image encoder by one Embodiment. (A)-(c) is a figure which shows the example of a restriction | limiting of the search range according to search mode, respectively. (A) And (b) is a figure which shows an example of the set of PU respectively applied with respect to encoding object CTU. It is a figure which shows an example of the relationship between the log | history of memory access amount, and target access amount. It is a figure which shows an example of distribution of the memory access amount for every search mode, and encoding cost. It is an operation | movement flowchart of a moving image encoding process. It is a figure which shows another example of distribution of the memory access amount for every search mode, and encoding cost. It is an operation | movement flowchart of the moving image encoding process by a modification. It is a block diagram of the computer which operate | moves as a moving image encoder by the computer program which implement | achieves the function of each part of the moving image encoder by embodiment or its modification.

  Hereinafter, the moving picture coding apparatus will be described with reference to the drawings. This moving image encoding device, when performing a motion search process for an encoding target block to which an inter prediction encoding mode is applied, uses a search mode to be applied from among a plurality of search modes that define a motion vector search method. Select. At this time, the moving image encoding apparatus performs a motion search process on a reduced picture generated by thinning out pixels from an encoded reference picture, so that the encoding of the encoding target block is performed for each of a plurality of search modes. An encoding cost that is an index of the amount is calculated. Further, the moving image encoding apparatus obtains a memory access amount representing a data amount read in the motion search process from the storage unit storing the reference picture for each search mode. Then, the moving image encoding apparatus selects a search mode with the minimum encoding cost as a search mode to be applied to the encoding target block from search modes having a memory access amount equal to or less than a target memory access amount. To do. Thereby, this moving image encoding device reduces the memory access amount in the motion search process while suppressing a decrease in encoding efficiency.

  In the present embodiment, the moving image encoding device encodes moving image data in conformity with HEVC. First, a method for dividing a picture into a plurality of blocks according to HEVC will be described.

FIG. 1 is a diagram illustrating an example of picture division by HEVC.
As shown in FIG. 1, a picture 100 is divided in units of Coding Tree Units (CTUs) that are examples of blocks, and each CTU 101 is encoded in the order of raster scanning. The size of the CTU 101 can be selected from 16x16 to 64x64 pixels. However, the size of the CTU 101 is constant for each sequence.

  The CTU 101 is further divided into a plurality of Coding Units (CU) 102 in a quadtree structure. Each CU 102 in one CTU 101 is encoded in the Z scan order. The size of the CU 102 is variable, and the size is selected from 8 × 8 to 64 × 64 pixels in the CU division mode. The CU 102 is a unit for selecting an intra prediction encoding mode and an inter prediction encoding mode that are encoding modes. The CU 102 is individually processed in units of Prediction Unit (PU) 103 or Transform Unit (TU) 104. The PU 103 is a unit for performing prediction according to the encoding mode. For example, in the intra prediction coding mode, the PU 103 is a unit to which a prediction mode indicating a reference direction for obtaining a value of each pixel in the prediction block is applied when generating a prediction block, and in the inter prediction coding mode. This is a unit for performing motion compensation. The size of the PU 103 can be selected, for example, from PU partition modes PartMode = 2Nx2N, NxN, 2NxN, Nx2N, 2NxU, 2NxnD, nRx2N, and nLx2N in inter prediction coding.

  On the other hand, the TU 104 is a unit of orthogonal transformation. In the intra prediction encoding mode, the TU 104 is also a prediction block generation unit. The size of the TU 104 is selected from 4 × 4 pixels to 32 × 32 pixels. The TU 104 is divided by a quadtree structure and processed in the Z scan order. Each of CU, PU, and TU is an example of a sub-block.

  FIG. 2 is a schematic configuration diagram of the moving picture encoding apparatus according to the present embodiment. The moving image encoding apparatus 1 includes an encoding cost calculation unit 11, a memory access amount calculation unit 12, a target access amount determination unit 13, a search mode determination unit 14, a motion vector calculation unit 15, and an encoding mode determination. Unit 16, encoding unit 17, and storage unit 18.

  Each of these units included in the moving image encoding apparatus 1 is formed as a separate circuit. Alternatively, these units included in the video encoding device 1 may be mounted on the video encoding device 1 as one integrated circuit in which circuits corresponding to the respective units are integrated. Furthermore, each of these units included in the moving image encoding device 1 may be a functional module realized by a computer program executed on a processor included in the moving image encoding device 1.

  A picture to be encoded is divided into a plurality of CTUs, for example, by a control unit (not shown) that controls the entire moving image encoding apparatus 1. Each CTU is input to the moving image encoding apparatus 1 in the raster scan order, for example. Then, the moving image encoding apparatus 1 performs encoding for each CTU. Hereinafter, each part which the moving image encoder 1 has is demonstrated.

  The processing of the encoding cost calculation unit 11, the memory access amount calculation unit 12, the target access amount determination unit 13, the search mode determination unit 14, and the motion vector calculation unit 15 applies an inter prediction encoding mode such as P picture or B picture. Performed on possible pictures. For blocks that are bidirectionally predicted, such as a CTU on a B picture, for each prediction direction, an encoding cost calculation unit 11, a memory access amount calculation unit 12, a target access amount determination unit 13, and a search mode determination unit. 14 and the motion vector calculation unit 15 are executed.

  For each of the plurality of search modes, the encoding cost calculation unit 11 encodes a motion vector candidate for specifying a motion vector search range for each PU that can be set as the encoding target CTU and an index of the code amount of the CTU. Calculate the cost.

  In the present embodiment, in order to reduce the amount of calculation required for motion vector candidate calculation and encoding cost calculation, the encoding cost calculation unit 11 extracts one pixel for each predetermined number of pixels from the reference picture. The generated reduced picture is read from the storage unit 18 and used.

  Furthermore, the encoding cost calculation unit 11 extracts one pixel for each predetermined number of pixels from the encoding target CTU and generates a reduced CTU. For example, the encoding cost calculation unit 11 extracts a pixel every other pixel from the CTU to be encoded and generates a reduced CTU in each of the horizontal direction and the vertical direction. When the size of the target PU is 8 pixels × 8 pixels or more, the encoding cost calculation unit 11 performs one encoding for each group of four consecutive pixels of the encoding target CTU in the horizontal direction and the vertical direction. Pixels may be extracted to generate a reduced CTU.

In order to reduce the memory access amount to the storage unit 18 in the motion search process by the motion vector calculation unit 15, the encoding cost calculation unit 11 sets, for example, a search mode with a limited search range as described below.
(A) Search mode 1:
In this search mode, the search range set for each PU is the widest. For example, the search range is set on the reference picture as a region obtained by combining an area having the same size as the target PU and an offset area having a predetermined width along the outer periphery thereof. Furthermore, as a region corresponding to the search range, which is referred to for determining a motion vector candidate, on the reduced picture, a region having the same position and size as the target PU, and a predetermined width along the outer periphery thereof (for example, An area obtained by combining offset areas having 3 to 5 pixels) is set. In this search mode, the encoding cost calculation unit 11 can refer to a plurality of reduced pictures according to the applied motion vector setting method.

In addition, about each search mode demonstrated below, a different point from the search mode 1 is demonstrated.
(B) Search mode 2:
In this search mode, the encoding cost calculation unit 11 prohibits the application of the minimum size PU. FIG. 3A is a diagram illustrating an example of a search range restriction for the search mode 2. The ratio of the offset area in the search range 311 for the PU 310 applied to the encoding target CTU 300 is smaller than the ratio of the offset area in the search range 321 for the PU 320 smaller than the PU 310. Thus, as the PU size is smaller, the ratio of the offset area to the search range increases. Therefore, the total number of pixels included in each search range when the encoding target CTU 300 is divided into a plurality of PUs 310 is the number of pixels included in each search range when the encoding target CTU 300 is divided into a plurality of PUs 320. More than the sum of Therefore, by prohibiting application of the minimum PU size, the memory access amount in the motion search process in the motion vector calculation unit 15 is reduced.
(C) Search mode 3:
In this search mode, the encoding cost calculation unit 11 prohibits application of the smallest size PU and the next largest size PU after the smallest size.

(D) Search mode 4:
In this search mode, the encoding cost calculation unit 11 narrows the width of the offset region on the reference picture for setting the search range by one pixel.
FIG. 3B is a diagram illustrating an example of a search range restriction for the search mode 4. In contrast to the width of the offset region 331 in the search mode 1 for the PU 330 applicable to the encoding target CTU, in the search mode 4, the width becomes narrower like the offset region 332. Therefore, the search range 334 in the search mode 4 is narrower than the search range 333 in the search mode 1. Therefore, the memory access amount in the motion search process in the motion vector calculation unit 15 is reduced.
(E) Search mode 5:
In this search mode, the encoding cost calculation unit 11 narrows the width of the offset area on the reference picture for setting the search range by two pixels.
Note that the encoding cost calculation unit 11 sets an area on the reduced picture for searching for a motion vector candidate by an amount corresponding to the narrowing of the offset area on the reference picture in search modes 4 and 5. The width of the offset region may be narrowed.

(F) Search mode 6:
In this search mode, the encoding cost calculation unit 11 makes the number of reference pictures smaller than the number of reference pictures in search mode 1.
FIG. 3C is a diagram illustrating an example of a search range restriction for the search mode 6. The greater the number of reference pictures, the higher the possibility that different reference pictures will be referenced for each PU. For example, in search mode 1, it is assumed that the reference picture 361 has the minimum encoding cost for the PU 351 included in the encoding target CTU 350, and the reference picture 362 has the minimum encoding cost for the PU 352. Furthermore, it is assumed that the coding cost of the PU 353 and the PU 354 is the minimum for the reference picture 363. In contrast, in the search mode 6, it is assumed that the reference picture 361 is referred to for each of the PUs 351 to 354. In this case, in search mode 1, the reference pictures for PU 351 and PU 352 are different from the reference pictures for PU 353 and PU 354. Therefore, the search range 371 for PU 351 and the search range 372 for PU 352 do not overlap with the search range 373 for PU 353 and the search range 374 for PU 354. On the other hand, in search mode 6, since the reference pictures for each PU are the same, the search ranges 381 to 384 for each PU may overlap each other. Therefore, the memory access amount in the motion search process in the motion vector calculation unit 15 is reduced.

  The encoding cost calculation unit 11 may set a search mode other than the above. For example, the encoding cost calculation unit 11 may set a search mode that combines two or more of the search modes 2 to 6 described above. Alternatively, the encoding cost calculation unit 11 may set a search mode in which the total number of PUs included in the encoding target CTU is equal to or less than a predetermined number.

  For each search mode, the encoding cost calculation unit 11 performs block matching for each applicable PU in the area on the reduced picture corresponding to the search range, and the reduced picture that most closely matches the PU and its reduction Determine the position of the reference area on the picture. In the present embodiment, the encoding cost calculation unit 11 may use a block in the reduced CTU corresponding to the focused PU set in the encoding target CTU for block matching. Then, the encoding cost calculation unit 11 calculates a vector representing a spatial movement amount between the PU and its reference area as a motion vector candidate.

  For example, in block matching, the encoding cost calculation unit 11 calculates a pixel difference absolute value sum SAD between a block in the reduced CTU corresponding to the focused PU and an area having the same size as the block. Then, the coding cost calculation unit 11 sets a region where the pixel difference absolute value sum SAD is minimum in a region where a motion vector candidate is searched as a reference region of a focused PU.

  It is estimated that the smaller the pixel difference absolute value sum SAD, the smaller the value of each pixel included in the prediction block, resulting in a smaller code amount. Therefore, the encoding cost calculation unit 11 calculates, for each search mode, for each applicable PU combination, the sum ΣSAD of the pixel difference absolute value sum SAD corresponding to the motion vector candidates of each PU included in the combination. The encoding cost calculation unit 11 sets the minimum value of the total ΣSAD as the encoding cost of the search mode.

  FIGS. 4A and 4B are diagrams illustrating an example of a set of PUs applied to an encoding target CTU. In the example illustrated in FIG. 4A, the encoding target CTU 400 is equally divided into four PUs 401 to 404. Therefore, the encoding cost is the sum of the pixel difference absolute value sums SAD1 to SAD4 of each of the four PUs 401 to 404. Note that the reference pictures referenced by each PU may be different. However, when the reference picture is limited to one in the restriction mode 6, the reference picture referred to by each PU is the same. On the other hand, in the example shown in FIG. 4B, the encoding target CTU 400 is such that the lower right PU 404 is further divided into four PUs 405 to 408. Therefore, the encoding cost is the sum of the pixel difference absolute value sums SAD1 to SAD3 and SAD5 to SAD8 of each of the seven PUs 401 to 403 and 405 to 408.

The encoding cost calculation unit 11 calculates a motion vector candidate and a difference absolute value sum SAD in order from the PU with the smaller size in order to reduce the amount of calculation, and saves it in the memory included in the encoding cost calculation unit 11. May be. Then, the coding cost calculation unit 11 includes the motion vector candidate and the pixel difference of the target PU, based on the motion vector candidate and the pixel difference absolute value sum SAD of the PU for which motion vector candidates have already been obtained, included in the target PU. An absolute value sum SAD may be calculated. For example, when the size of the focused PU is 16 × 16 pixels, the encoding cost calculation unit 11 focuses on the average value of the motion vector candidates obtained for each of the four 8 × 8 pixel PUs included in the PU. It is good also as a motion vector candidate of PU to perform. Then, the encoding cost calculation unit 11 may set the sum of the pixel difference absolute value sums SAD of the four 8 × 8 pixel PUs included in the target PU as the pixel difference absolute value sum SAD of the target PU.
Also, the encoding cost calculation unit 11 obtains the pixel difference absolute value sum SAD for the PU that refers to the reduced picture corresponding to the reference picture referenced in the search mode 6 obtained in the search modes 1 to 5 in the search mode. 6 may be used. Thereby, the calculation amount in the search mode 6 is reduced.

  For each search mode, the encoding cost calculation unit 11 sets the encoding cost for the encoding target CTU, the combination of PUs to be applied, and the motion vector candidate obtained for each PU to the memory access amount calculation unit 12 and Output to the search mode determination unit 14.

  The memory access amount calculation unit 12 calculates a memory access amount representing the amount of data read from the storage unit 18 in the motion search process by the motion vector calculation unit 15 for each search mode for the encoding target CTU. In this embodiment, the memory access amount calculation unit 12 calculates the memory access amount according to the search mode and the search range indicated by the motion vector candidates of each PU in the search mode. For example, the memory access amount calculation unit 12 calculates the total number of pixels included in the search range indicated by each PU as the number of pixels read from the storage unit 18 for the search mode of interest. Depending on the positional relationship between the search ranges, a plurality of search ranges may overlap each other. In this case, the memory access amount calculation unit 12 only needs to count the pixels included in the overlapping area once. In addition, the size of the search range in the motion vector calculation unit 15 is obtained by combining an offset area having a predetermined width (for example, 6 to 10 pixels) along the outer periphery of the area in the same size as the PU size. This is the size of the area. Therefore, in the search mode in which the offset region is narrow as in the search mode 4 or the search mode 5, the search range is narrowed.

  The memory access amount calculation unit 12 calculates 100 as a value obtained by dividing the number of pixels read from the storage unit 18 by the number of pixels of the encoding target CTU (for example, 64 pixels × 64 pixels = 4096 pixels) for the search mode of interest. The percentage value obtained by multiplying is set as the memory access amount of the search mode. The memory access amount may be the number of pixels read from the storage unit 18 itself, or a value obtained by multiplying the number of pixels read from the storage unit 18 by the number of bits or the number of bytes per pixel. There may be. Then, the memory access amount calculation unit 12 outputs the memory access amount of each search mode to the search mode determination unit 14.

  The target access amount determination unit 13 determines a target access amount that is the upper limit of the memory access amount in the motion vector calculation unit 15 for each encoding target CTU. In the present embodiment, the target access amount determination unit 13 increases the target access amount as the total of the actual memory access amounts of the most recent predetermined number of encoded CTUs is smaller. For example, the target access amount determination unit 13 determines the actual value for the most recent (n−1) CTUs from a value obtained by multiplying the reference value of the memory access amount per CTU by n (where n is an integer of 2 or more). The target access amount is obtained by subtracting the total memory access amount. For example, it is assumed that n = 10 and the reference value of the memory access amount per CTU is 550 (%). Then, it is assumed that the total memory access amount to the storage unit 18 of the motion vector calculation unit 15 is 4900 (%) for the latest nine CTUs. In this case, the target access amount for the encoding target CTU is 600 (%) (= 550 × 10−4900). By setting the target access amount in this way, the video encoding device 1 can adaptively set the memory access amount for the encoding target CTU.

  Note that the reference value of the memory access amount per CTU is set according to, for example, hardware resources of the video encoding device 1 or a required encoding rate. However, the reference value is preferably set to a value smaller than the memory access amount when the encoding target CTU is divided by the PU of the minimum size and the search range is not limited for each PU. The reference value is preferably set to a value larger than the memory access amount when the encoding target CTU is divided by the maximum size PU and the search range is minimized for each PU. Also, n is set to any value within the range of the number of CTUs included in 2 to 1 picture, for example. As n increases, the target access amount determination unit 13 refers to the memory access amount for a large number of encoded CTUs, so that the increase or decrease in the target access amount for the encoding target CTU with respect to the reference value is likely to be allowed. .

  FIG. 5 is a diagram illustrating an example of the relationship between the history of the memory access amount and the target access amount. In FIG. 5, the horizontal axis represents the number of encoded CTUs, and the vertical axis represents the memory access amount. A line 500 represents a reference value of the memory access amount per CTU. A graph 501 represents a history of memory access amount for each CTU. For example, the target access amount for the mth CTU is determined for the memory access amount of the nearest (m− (n−1)) to (m−1) th CTU. Therefore, if the average value of the memory access amount of the (m− (n−1)) to (m−1) th CTU is lower than the reference value, the target access amount determination unit 13 sets the target for the mth CTU. The access amount can be made larger than the reference value. Conversely, if the average memory access amount of the (m- (n-1)) to (m-1) th CTU is higher than the reference value, the target access amount for the mth CTU is less than the reference value. Also lower.

  The target access amount determination unit 13 notifies the search mode determination unit 14 of the target access amount for each encoding target CTU. Further, the target access amount determination unit 13 determines the memory access amount of the search mode selected for the encoding target CTU notified from the search mode determination unit 14 in order to determine the target access amount for the next and subsequent CTUs. The data is stored in a memory included in the target access amount determination unit 13.

  The search mode determination unit 14 determines, for each encoding target CTU, a search mode to be applied from among the plurality of search modes based on the target access amount and the encoding cost of each of the plurality of search modes.

  In the present embodiment, the search mode determination unit 14 selects a search mode that applies a search mode that minimizes the coding cost among search modes in which the memory access amount is equal to or less than the target access amount.

  FIG. 6 is a diagram illustrating an example of a memory access amount and a coding cost distribution for each search mode. In FIG. 6, the horizontal axis represents the memory access amount, and the vertical axis represents the coding cost. Points 601 to 606 represent the memory access amount and encoding cost for search mode 1 to search mode 6, respectively. A line 610 represents the target access amount. In this example, the memory access amount in the search mode 1 represented by the point 601 is larger than the target access amount. Therefore, search mode 1 is not selected. Of the remaining search modes in which the memory access amount is equal to or less than the target access amount, the encoding cost of search mode 2 indicated by point 602 is minimized. Therefore, in this example, search mode 2 is selected.

  Note that the search mode determination unit 14 may select a search mode for each prediction direction when bidirectional prediction is possible, such as when the encoding target picture is a B picture. Alternatively, the search mode determination unit 14 may select the same search mode for all prediction directions. In this case, the memory access amount and the encoding cost for each search mode can be the sum of the memory access amount and the encoding cost obtained for each prediction direction.

  The search mode determination unit 14 outputs the selected search mode, the combination of PUs corresponding to the search mode, and the motion vector candidates of each PU to the motion vector calculation unit 15 for each encoding target CTU. In addition, the search mode determination unit 14 notifies the target access amount determination unit 13 of the selected search mode.

  The motion vector calculation unit 15 sets a search range for each PU based on the search mode and the motion vector candidates of each PU notified from the search mode determination unit 14 for the encoding target CTU, and moves within the search range. Calculate the vector.

  In the present embodiment, the motion vector calculation unit 15 identifies an area that has moved by the number of pixels in the horizontal and vertical directions indicated by the motion vector candidate from the area at the same position as the PU on the reference picture for the target PU. . Then, the motion vector calculation unit 15 sets an area obtained by expanding the specified area by an offset area having a predetermined width (for example, 6 pixels to 10 pixels) along the outer periphery. At this time, when a search mode in which the width of the offset area is limited is set as in search mode 4 or 5, the width of the offset area is reduced according to the restriction. For example, in search mode 4, the motion vector calculation unit 15 narrows the width of the offset region by one pixel. Similarly, in search mode 5, the motion vector calculation unit 15 narrows the width of the offset region by two pixels.

  The motion vector calculation unit 15 reads the value of the pixel included in the search range from the storage unit 18 for each PU in the encoding target CTU. In that case, when the search range of several PU overlaps, the motion vector calculation part 15 may read from the memory | storage part 18 about the pixel contained in the duplication part only once.

  The motion vector calculation unit 15 performs block matching within the search range for each PU in the encoding target CTU, and determines the position of the region within the search range that most closely matches the PU. Then, the motion vector calculation unit 15 calculates a vector representing the amount of spatial movement between the PU and its area as a motion vector. Note that the motion vector calculation unit 15 performs an interpolation process on the search range so that the accuracy of the motion vector is less than one pixel, and ½ pixel or 1/1 for each pixel in the search range. An interpolation search range having pixels whose positions are shifted in the horizontal direction or the vertical direction by 4 pixels may be set. Then, the motion vector calculation unit 15 may calculate a motion vector by performing block matching on each of the search range and the interpolation search range to obtain the position of the region that most closely matches the PU.

  The motion vector calculation unit 15 outputs information indicating the motion vector of each PU in the encoding target CTU and the reference picture referred to by the motion vector to the encoding mode determination unit 16.

  The encoding mode determination unit 16 obtains a combination of applicable CU size, PU size, TU size, and encoding mode in the encoding target CTU. Then, the encoding mode determination unit 16 calculates an encoding cost that is an estimated value of the code amount for each combination. Then, the encoding mode determination unit 16 sets the combination that minimizes the encoding cost as the CU size, PU size, TU size, and encoding mode to be applied to the encoding target CTU.

  In the present embodiment, when setting the combination, the encoding mode determination unit 16 sets the motion vector for each PU set according to the selected search mode when the encoding mode is the inter prediction encoding mode. Is used. That is, the encoding mode determination unit 16 generates a prediction block by performing motion compensation on the reference region on the reference picture with the motion vector determined according to the selected search mode.

For example, in order to calculate the encoding cost of the combination of interest, the encoding mode determination unit 16 calculates a prediction error, that is, a pixel difference absolute value sum SAD for each TU included in the combination according to the following equation.
SAD = Σ | OrgPixel-PredPixel |
Here, OrgPixel is the value of the pixel of interest in the encoding target picture, for example, the value of the pixel included in the TU, and PredPixel is the value of the pixel included in the prediction block corresponding to the block of interest, which is obtained in the HEVC standard . A prediction block is generated from a coded reference picture or another coded block according to a coding mode (intra prediction coding mode or inter prediction coding mode) included in the combination of interest.

  The encoding mode determination unit 16 may calculate the absolute value sum SATD of each pixel after Hadamard transform of the difference image between the block of interest and the prediction block, instead of SAD.

The encoding mode determination unit 16 calculates the encoding cost Cost according to the following expression for the combination of interest.
Cost = ΣSAD + λR
Here, ΣSAD is the sum of SADs calculated for each TU included in the CTU to be encoded. R is an estimated value of code amount for items other than orthogonal transform coefficients, such as a motion vector and a flag indicating a prediction mode indicating a prediction direction in the intra prediction encoding mode. Λ is a constant.

  When the encoding mode determination unit 16 obtains a combination that minimizes the encoding cost, the encoding unit determines the CU size, the PU size, the TU size, and the encoding mode that are included in the combination and are applied to the CTU to be encoded. 17 to output. Note that, for a CU to be inter-predictively encoded, the encoding mode determination unit 16 also outputs a motion vector calculated for a PU included in the CU to the encoding unit 17.

  The encoding unit 17 divides the CTU to be encoded according to the CU size, PU size, and TU size determined by the encoding mode determination unit 16. Then, the encoding unit 17 creates a prediction block for each TU included in the CU according to the encoding mode for each CU determined by the encoding mode determination unit 16. Then, the encoding unit 17 quantizes and variable-length-codes the orthogonal transform coefficient obtained by orthogonally transforming the prediction error signal between the CTU to be encoded and the prediction block for each TU, thereby obtaining the encoding target. Encode the CTU.

  Referring to FIG. 2 again, the encoding unit 17 includes a prediction block generation unit 21, a prediction error signal calculation unit 22, an orthogonal transformation unit 23, a quantization unit 24, a decoding unit 25, and a variable length encoding unit. 26.

  The prediction block generation unit 21 generates a prediction block. When the target CU included in the CTU to be encoded is inter-predictively encoded, the prediction block generation unit 21 uses the motion vector set in the PU for each PU included in the CU to generate a reference picture. A prediction block is generated by performing motion compensation. In addition, when the target CU is subjected to intra-prediction encoding, the prediction block generation unit 21 encodes blocks around the PU according to the prediction mode applied to the PU for each PU included in the CU. A prediction block is generated from the pixels included in.

  The prediction block generation unit 21 outputs the prediction block to the prediction error signal calculation unit 22.

  The prediction error signal calculation unit 22 performs a difference calculation for each pixel in the CTU to be encoded with the corresponding pixel of the prediction block. And the prediction error signal calculation part 22 makes the difference value corresponding to each pixel obtained by the difference calculation a prediction error signal. The prediction error signal calculation unit 22 outputs the prediction error signal to the orthogonal transformation unit 23.

The orthogonal transform unit 23 obtains a set of orthogonal transform coefficients by performing orthogonal transform on the prediction error signal for each TU in the CTU to be encoded. For example, the orthogonal transform unit 23 obtains a set of DCT coefficients for each TU as orthogonal transform coefficients by using discrete cosine transform (DCT) as orthogonal transform processing.
The orthogonal transform unit 23 outputs a set of orthogonal transform coefficients for each TU to the quantization unit 24.

  For each TU, the quantization unit 24 quantizes the orthogonal transform coefficient to calculate a quantization coefficient of the orthogonal transform coefficient. This quantization process is a process that represents a signal value included in a certain section as one signal value. The fixed interval is called a quantization width. For example, the quantization unit 24 quantizes the orthogonal transform coefficient by truncating a predetermined number of lower bits corresponding to the quantization width from the orthogonal transform coefficient. The quantization width is determined by the quantization parameter. For example, the quantization unit 24 determines a quantization width to be used according to a function representing a quantization width value with respect to a quantization parameter value. The function can be a monotonically increasing function with respect to the value of the quantization parameter, and is set in advance.

Further, the quantization unit 24 may determine the quantization parameter according to any of various quantization parameter determination methods corresponding to a moving image coding standard such as HEVC. The quantization unit 24 can use, for example, a quantization parameter calculation method related to the MPEG-2 standard test model 5. For the quantization parameter calculation method for the MPEG-2 standard test model 5, refer to the URL specified at http://www.mpeg.org/MPEG/MSSG/tm5/Ch10/Ch10.html, for example. I want to be.
Since the quantization unit 24 can reduce the number of bits used to represent the orthogonal transform coefficient by executing the quantization process, the amount of information included in the CTU to be encoded can be reduced. The quantization unit 24 outputs the quantized coefficient to the decoding unit 25 and the variable length coding unit 26.

  The decoding unit 25 is referred from the quantization coefficient of the CTU to be encoded in order to encode a picture subsequent to the CTU after the CTU and a picture including the CTU to be encoded in the encoding order. Generate a reference picture. For this purpose, the decoding unit 25 dequantizes the quantization coefficient by multiplying the quantization coefficient by a predetermined number corresponding to the quantization width determined by the quantization parameter. By this inverse quantization, a set of orthogonal transform coefficients of each TU, for example, a set of DCT coefficients is restored. Thereafter, the decoding unit 25 performs inverse orthogonal transform processing on the set of orthogonal transform coefficients for each TU. For example, when the orthogonal transform unit 23 uses DCT, the decoding unit 25 performs inverse DCT processing on each TU. By performing the inverse quantization process and the inverse orthogonal transform process on the quantized signal, a prediction error signal having the same level of information as the prediction error signal before encoding is reproduced.

The decoding unit 25 adds the reproduced prediction error signal corresponding to the pixel to each pixel value of the prediction block. By executing these processes for each prediction block, the decoding unit 25 restores a block that is referred to in order to generate a prediction block for a PU to be encoded thereafter. Further, the decoding unit 25 may perform a deblocking filter process on the restored block.
Each time the decoding unit 25 restores a block, the decoding unit 25 stores the restored block in the storage unit 18.

  The decoding unit 25 further generates a thinned block having a smaller number of pixels than the number of pixels of the block by thinning out the pixels from the restored block, and stores the thinned block in the storage unit 18. This decimation block is used to generate a reduced picture of the reference picture.

  The storage unit 18 temporarily stores the restored block. By combining the restored blocks for one picture according to the coding order of each block, a picture to be referred to when coding the subsequent picture is obtained. Similarly, a reduced picture corresponding to the reference picture is obtained by combining the thinned blocks for one picture according to the coding order of each block. The storage unit 18 stores a predetermined number of reference pictures and reduced pictures that may be referred to by the encoding target picture, and when the number of stored reference pictures exceeds the predetermined number, Discard the oldest reference picture and reduced picture in order.

  Further, the storage unit 18 stores a motion vector for each of the PUs subjected to inter prediction encoding.

  The variable length coding unit 26 performs variable length coding on the quantization coefficient included in the CTU to be coded. Furthermore, the variable length coding unit 26 also performs variable length coding on a motion vector or the like used to create a prediction block. Then, the variable length coding unit 26 outputs a bit stream in which coded bits obtained by the variable length coding are arranged in a predetermined order according to HEVC or the like. The variable length coding unit 26 uses a Huffman coding process such as Context-based Adaptive Variable Length Coding (CAVLC) or an arithmetic coding process such as Context-based Adaptive Binary Arithmetic Coding (CABAC) as the variable length coding method. be able to.

  A control unit (not shown) combines the output bit streams in a predetermined order and adds header information according to HEVC or the like to obtain a bit stream including encoded moving image data.

  FIG. 7 is an operation flowchart of the moving image encoding process performed by the moving image encoding device 1. The moving image encoding apparatus 1 performs encoding for each CTU according to the following operation flowchart.

  The control unit (not shown) determines whether or not the encoding target picture including the encoding target CTU is a P picture or a B picture (step S101). If the encoding target picture is a P picture or a B picture (step S101—Yes), the encoding cost calculation unit 11 determines the encoding cost of the encoding target CTU, the combination of PUs to be applied, and each PU for each search mode. Motion vector candidates are calculated (step S102). Then, the encoding cost calculation unit 11 outputs the encoding cost of the encoding target CTU, the combination of PUs, and motion vector candidates of each PU to the memory access amount calculation unit 12 and the search mode determination unit 14 for each search mode.

  The memory access amount calculation unit 12 calculates the memory access amount based on the search range of each PU for each search mode (step S103). Then, the memory access amount calculation unit 12 outputs the memory access amount of each search mode to the search mode determination unit 14.

  Further, the target access amount determination unit 13 determines the target access amount of the encoding target CTU based on the memory access amount of the latest (n−1) CTUs (step S104). Then, the target access amount determination unit 13 outputs the target access amount to the search mode determination unit 14.

  The search mode determination unit 14 selects a search mode that applies a search mode that minimizes the coding cost among search modes in which the memory access amount is equal to or less than the target access amount (step S105). Then, the search mode determination unit 14 outputs the selected search mode, the combination of PUs corresponding to the search mode, and the motion vector candidates of each PU to the motion vector calculation unit 15.

  The motion vector calculation unit 15 sets a search range for each PU based on the selected search mode and the motion vector candidates of each PU in the encoding target CTU for the search mode for the encoding target CTU, A motion vector is calculated within the search range (step S106).

  After step S106 or when the encoding target picture is an I picture in step S101 (step S101-No), the encoding mode determination unit 16 determines the encoding mode, CU size, Determine the combination of PU size and TU size. That is, the encoding mode determination unit 16 sets a combination that applies the minimum encoding cost to the encoding target CTU among combinations of the encoding mode, CU size, PU size, and TU size (step S107). . The prediction block generation unit 21 of the encoding unit 17 calculates a prediction block according to the encoding mode included in the determined combination (step S108). The prediction block generation unit 21 corresponds to the PU in the reference picture with the motion vector calculated by the motion vector calculation unit 15 for each PU in the CU that is inter-predictively encoded in the encoding target CTU. The prediction block is obtained by motion compensation of the reference region to be performed. And the prediction error signal calculation part 22 of the encoding part 17 calculates a prediction error signal by calculating the difference for every pixel between TU and the prediction block corresponding to the TU for every TU (step S109).

  Thereafter, the orthogonal transform unit 23 of the encoding unit 17 performs orthogonal transform on the prediction error signal for each TU to calculate a set of orthogonal transform coefficients (step S110). Then, the quantization unit 24 of the encoding unit 17 quantizes each orthogonal transform coefficient (step S111).

  The decoding unit 25 of the encoding unit 17 adds the prediction error signal reproduced by inverse quantization and inverse orthogonal transform of the quantized coefficient and the prediction block to generate a reference block (step S112). Then, the decoding unit 25 stores the reference block in the storage unit 18. Further, the decoding unit 25 also stores a reduced block generated by thinning out pixels from the reference block in the storage unit 18. On the other hand, the variable length coding unit 26 of the coding unit 17 performs variable length coding on each quantization coefficient (step S113). Then, the moving image encoding apparatus 1 ends the moving image encoding process.

  As described above, this moving picture encoding apparatus uses a reduced picture having a smaller number of pixels than the original picture, and for each CTU, the memory access amount is a target value from a plurality of search modes. The search mode with the lowest coding cost is selected from among the following. Therefore, this moving image encoding apparatus can suppress the memory access amount to the storage unit storing the reference picture at the time of calculating the motion vector while maintaining the encoding efficiency.

  According to the modification, the search mode determination unit 14 applies a search mode in which the memory access amount is minimum among search modes in which the encoding cost is equal to or less than a predetermined reference value to the encoding target CTU. The mode may be selected. In this modification, the moving picture encoding apparatus can select a search mode that can minimize the memory access amount while maintaining the encoding efficiency at a constant level. In this case, the target access amount determination unit 13 may be omitted.

  According to another modification, the search mode determination unit 14 performs encoding between two consecutive CTUs so that the image quality does not change suddenly between adjacent CTUs on the reproduced picture. The search mode may be selected so that the cost difference is equal to or less than the allowable change amount. In this case, the search mode determination unit 14 does not have a search mode in which the difference from the encoding cost of the immediately preceding CTU is equal to or less than the allowable change amount among search modes in which the memory access amount is equal to or less than the target access amount. There is. In such a case, the search mode determination unit 14 may select a search mode from search modes exceeding the target access amount. The allowable change amount can be, for example, the encoding cost for the latest CTU. That is, if the encoding cost of the encoding target CTU for a certain search mode is less than or equal to twice the encoding cost for the nearest CTU, the difference in encoding cost for that search mode is less than the allowable change amount.

  FIG. 8 is a diagram illustrating another example of the memory access amount and the encoding cost distribution for each search mode. In FIG. 8, the horizontal axis represents the memory access amount, and the vertical axis represents the coding cost. Points 801 to 806 represent the memory access amount and encoding cost for search mode 1 to search mode 6 for the encoding target CTU, respectively. A line 810 represents the target access amount. Furthermore, a line 820 represents the encoding cost for the immediately preceding CTU, and a line 821 represents the encoding cost obtained by adding an allowable change amount to the encoding cost for the immediately preceding CTU. In this example, in search mode 3 to search mode 6 in which the memory access amount is equal to or less than the target access amount, the change amount of the coding cost from the immediately preceding CTU becomes larger than the allowable change amount. Therefore, the search mode determination unit 14 selects the memory among the search mode 1 and the search mode 2 in which the change amount of the coding cost from the immediately preceding CTU is within the allowable change amount although the memory access amount is larger than the target access amount. The search mode 2 with the smaller access amount is selected.

  FIG. 9 is an operation flowchart of a moving image encoding process according to this modification. The moving image encoding apparatus 1 executes the process of each step shown in FIG. 9 instead of step S105 in the moving image encoding process shown in FIG.

After step S104, the search mode determination unit 14 determines whether there is a search mode in which the memory access amount is equal to or less than the target access amount and the difference between the encoding cost and the encoding cost of the immediately preceding CTU is equal to or less than the allowable change amount. It is determined whether or not (step S201). When there is a search mode that satisfies the condition of step S201 (step S201-Yes), the search mode determination unit 14 selects a search mode that minimizes the coding cost among the search modes that satisfy the condition (step S202). . On the other hand, when there is no search mode that satisfies the condition of step S201 (step S201-No), the search mode determination unit 14 searches the search mode in which the difference between the encoding cost and the encoding cost of the immediately preceding CTU is less than the allowable change amount. To extract. Then, the search mode determination unit 14 selects a search mode that minimizes the memory access amount from the extracted search modes (step S203).
After step S202 or S203, the moving image encoding apparatus 1 executes the processes after step S106.

  According to this modification, the moving picture coding apparatus can also suppress the memory access amount in the motion search while suppressing the sudden change in the quality of the reproduced picture between adjacent CTUs.

  According to another modification, when the memory access amount is not equal to or less than the target access amount for any of the plurality of search modes for the encoding target CTU, the search mode determination unit 14 performs intra processing for the encoding target CTU. It may be determined that the predictive coding mode is applied. Alternatively, for any search mode, even when the difference between the encoding cost for the most recent CTU and the encoding cost of the encoding target CTU is not less than or equal to the allowable change amount, the search mode determination unit 14 may be set to intra It may be determined that the predictive coding mode is applied. Then, the search mode determination unit 14 may notify the encoding mode determination unit 16 to that effect. In this case, the encoding mode determination unit 16 only needs to calculate the encoding cost when applying the intra prediction encoding mode for each combination of CU, PU, and TU applied to the encoding target CTU. .

  In this case, since the motion search processing in the motion vector calculation unit 15 is not performed, the memory access amount for the encoding target CTU by the motion vector calculation unit 15 becomes zero. Therefore, the search mode determination unit 14 can increase the target access amount for the subsequent CTU. As a result, with respect to the subsequent CTU, the degree of freedom in selecting the search mode is increased.

  According to still another modification, the motion vector calculation unit 15 calculates the PU combination to be applied to the encoding target CTU and the motion vector of each PU from the beginning according to the search mode selected by the search mode determination unit 14. May be executed. In this case, since the motion vector calculation unit 15 does not use the PU combination and the motion vector candidate obtained using the reduced picture, the motion vector increases, but the motion vector that further reduces the coding cost is obtained. Could be possible.

  According to still another modification, the video encoding device may conform to a video encoding method other than HEVC. In this case, the video encoding apparatus may set one motion vector for one encoding target block.

  FIG. 10 is a configuration diagram of a computer that operates as a moving image encoding apparatus by operating a computer program that realizes the functions of the respective units of the moving image encoding apparatus according to each of the above-described embodiments or modifications thereof.

  The computer 100 includes a user interface unit 101, a communication interface unit 102, a storage unit 103, a storage medium access device 104, and a processor 105. The processor 105 is connected to the user interface unit 101, the communication interface unit 102, the storage unit 103, and the storage medium access device 104 via, for example, a bus.

  The user interface unit 101 includes, for example, an input device such as a keyboard and a mouse and a display device such as a liquid crystal display. Alternatively, the user interface unit 101 may include a device such as a touch panel display in which an input device and a display device are integrated. Then, the user interface unit 101 outputs, to the processor 105, an operation signal for selecting moving image data to be encoded or encoded moving image data to be decoded in accordance with a user operation, for example. The user interface unit 101 may display the decoded moving image data received from the processor 105.

  The communication interface unit 102 may include a communication interface for connecting the computer 100 to a device that generates moving image data, for example, a video camera, and a control circuit thereof. Such a communication interface can be, for example, Universal Serial Bus (Universal Serial Bus, USB).

  Furthermore, the communication interface unit 102 may include a communication interface for connecting to a communication network according to a communication standard such as Ethernet (registered trademark) and a control circuit thereof.

  In this case, the communication interface unit 102 acquires moving image data to be encoded from another device connected to the communication network, and passes the data to the processor 105. Further, the communication interface unit 102 may output the encoded moving image data received from the processor 105 to another device via a communication network.

  The storage unit 103 includes, for example, a readable / writable semiconductor memory and a read-only semiconductor memory. The storage unit 103 stores a computer program for executing a moving image encoding process executed on the processor 105, and data generated during or as a result of these processes.

  The storage medium access device 104 is a device that accesses a storage medium 106 such as a magnetic disk, a semiconductor memory card, and an optical storage medium. For example, the storage medium access device 104 reads a computer program for moving image encoding processing executed on the processor 105 stored in the storage medium 106 and passes the computer program to the processor 105.

  The processor 105 generates encoded moving image data by executing the computer program for moving image encoding processing according to the above-described embodiment or modification. The processor 105 stores the generated encoded moving image data in the storage unit 103 or outputs it to another device via the communication interface unit 102.

  Note that the computer program capable of executing the functions of the respective units of the moving image encoding device 1 on the processor may be provided in a form recorded on a computer-readable medium. However, such a recording medium does not include a carrier wave.

  All examples and specific terms listed herein are intended for instructional purposes to help the reader understand the concepts contributed by the inventor to the present invention and the promotion of the technology. It should be construed that it is not limited to the construction of any example herein, such specific examples and conditions, with respect to showing the superiority and inferiority of the present invention. Although embodiments of the present invention have been described in detail, it should be understood that various changes, substitutions and modifications can be made thereto without departing from the spirit and scope of the present invention.

The following supplementary notes are further disclosed regarding the embodiment described above and its modifications.
(Appendix 1)
A video encoding device that encodes an error between a block to be encoded on a picture included in video data and a reference block indicated by a motion vector on another encoded picture,
For each of a plurality of search modes defining the motion vector search method, it corresponds to the search range of the motion vector defined in the search mode on the reduced picture obtained by reducing the number of pixels included in the other picture. A motion vector candidate for obtaining an area that most closely matches a reduced block obtained by reducing the number of pixels included in the encoding target block within the area, and specifying the search range from a movement amount between the reduced block and the most coincident area And an encoding cost calculation unit for calculating an encoding cost representing a code amount of the encoding target block;
For each of the plurality of search modes, based on the search range in the search mode, a memory access amount representing a data amount to be read from the storage unit storing the other picture to calculate the motion vector A memory access amount calculation unit to calculate,
A search mode determination unit that selects a search mode to be applied to the encoding target block from the plurality of search modes based on the memory access amount and the encoding cost for each of the plurality of search modes;
According to the selected search mode, the search range is set on the other picture, the pixel data included in the search range on the other picture is read from the storage unit, and the code is read within the search range. A motion vector calculation unit that obtains a region that most closely matches the encoding target block and calculates a vector representing a movement amount between the encoding target block and the region as the motion vector;
A moving picture encoding apparatus having:
(Appendix 2)
The encoding target block can be divided into a plurality of sub-blocks having different sizes,
The encoding cost calculation unit, for each of the plurality of search modes, for each combination of the subblocks, the encoding cost representing the motion vector candidate of each subblock included in the combination and the code amount of the subblock. Is determined as the encoding cost of the encoding target block, and the combination of the subblocks corresponding to the minimum value is a combination of subblocks for the search mode. age,
The memory access amount calculation unit, for each of the plurality of search modes, based on the positional relationship of the search range indicated by the motion vector candidate of each subblock included in the combination of the subblocks for the search mode Obtaining the memory access amount for the search mode;
The moving image encoding apparatus according to appendix 1, wherein the motion vector calculation unit calculates the motion vector for each sub-block included in the combination of the sub-blocks for the selected search mode.
(Appendix 3)
The plurality of search modes include a first search mode in which a size of the search range for each subblock is a first size, and a size of the search range for each subblock is smaller than the first size. The moving image encoding apparatus according to attachment 2, including a second search mode having a second size.
(Appendix 4)
The plurality of search modes further include a third search mode in which a minimum value of the applicable size of the sub-block is larger than a minimum value of the applicable size of the sub-block in the first search mode. The moving picture encoding apparatus according to attachment 3.
(Appendix 5)
The moving image encoding apparatus according to Supplementary Note 3 or 4, wherein the plurality of search modes further include a fourth search mode in which the number of other pictures that can be referred to is smaller than that in the first search mode. (Figure 3)
(Appendix 6)
The search mode determination unit selects a search mode in which the coding cost is the lowest among search modes in which the memory access amount is equal to or less than a target access amount among the plurality of search modes. The moving image encoding device according to any one of claims.
(Appendix 7)
The moving image encoding apparatus according to appendix 6, further comprising a target access amount determination unit that increases the target access amount as the total of the memory access amounts of each of the predetermined number of blocks closest to the encoding target block decreases. .
(Appendix 8)
The search mode determination unit, for any search mode in which the memory access amount is equal to or less than the target access amount among the plurality of search modes, the coding cost of the search mode and the block immediately before the block. The moving image according to appendix 6 or 7, wherein when the difference from the encoding cost is out of an allowable range, a search mode in which the memory access amount is minimum is selected from search modes in which the difference is included in the allowable range. Image encoding device.
(Appendix 9)
A video encoding method for encoding an error between a block to be encoded on a picture included in video data and a reference block indicated by a motion vector on another encoded picture,
For each of a plurality of search modes defining the motion vector search method, it corresponds to the search range of the motion vector defined in the search mode on the reduced picture obtained by reducing the number of pixels included in the other picture. A motion vector candidate for obtaining an area that most closely matches a reduced block obtained by reducing the number of pixels included in the encoding target block within the area, and specifying the search range from a movement amount between the reduced block and the most coincident area And an encoding cost representing the code amount of the encoding target block,
For each of the plurality of search modes, based on the search range in the search mode, a memory access amount representing a data amount to be read from the storage unit storing the other picture to calculate the motion vector Calculate
Based on the memory access amount and the encoding cost for each of the plurality of search modes, select a search mode to be applied to the encoding target block from the plurality of search modes,
According to the selected search mode, the search range is set on the other picture, the pixel data included in the search range on the other picture is read from the storage unit, and the code is read within the search range. A region that most closely matches the encoding target block is calculated, and a vector representing a movement amount between the encoding target block and the region is calculated as the motion vector.
A moving picture encoding method including the above.
(Appendix 10)
A moving picture coding computer program for coding an error between a coding target block on a picture included in moving picture data and a reference block indicated by a motion vector on another coded picture. And
For each of a plurality of search modes defining the motion vector search method, it corresponds to the search range of the motion vector defined in the search mode on the reduced picture obtained by reducing the number of pixels included in the other picture. A motion vector candidate for obtaining an area that most closely matches a reduced block obtained by reducing the number of pixels included in the encoding target block within the area, and specifying the search range from a movement amount between the reduced block and the most coincident area And an encoding cost representing the code amount of the encoding target block,
For each of the plurality of search modes, based on the search range in the search mode, a memory access amount representing a data amount to be read from the storage unit storing the other picture to calculate the motion vector Calculate
Based on the memory access amount and the encoding cost for each of the plurality of search modes, select a search mode to be applied to the encoding target block from the plurality of search modes,
According to the selected search mode, the search range is set on the other picture, the pixel data included in the search range on the other picture is read from the storage unit, and the code is read within the search range. A region that most closely matches the encoding target block is calculated, and a vector representing a movement amount between the encoding target block and the region is calculated as the motion vector.
A computer program for encoding a moving image for causing a computer to execute the above.

DESCRIPTION OF SYMBOLS 1 Moving image encoder 11 Coding cost calculation part 12 Memory access amount calculation part 13 Target access amount determination part 14 Search mode determination part 15 Motion vector calculation part 16 Encoding mode determination part 17 Encoding part 18 Storage part 21 Prediction block Generation unit 22 Prediction error signal calculation unit 23 Orthogonal transformation unit 24 Quantization unit 25 Decoding unit 26 Variable length coding unit 100 Computer 101 User interface unit 102 Communication interface unit 103 Storage unit 104 Storage medium access device 105 Processor 106 Storage medium

Claims (7)

A video encoding device that encodes an error between a block to be encoded on a picture included in video data and a reference block indicated by a motion vector on another encoded picture,
For each of a plurality of search modes defining the motion vector search method, it corresponds to the search range of the motion vector defined in the search mode on the reduced picture obtained by reducing the number of pixels included in the other picture. A motion vector candidate for obtaining an area that most closely matches a reduced block obtained by reducing the number of pixels included in the encoding target block within the area, and specifying the search range from a movement amount between the reduced block and the most coincident area And an encoding cost calculation unit for calculating an encoding cost representing a code amount of the encoding target block;
For each of the plurality of search modes, based on the search range in the search mode, a memory access amount representing a data amount to be read from the storage unit storing the other picture to calculate the motion vector A memory access amount calculation unit to calculate,
A search mode determination unit that selects a search mode to be applied to the encoding target block from the plurality of search modes based on the memory access amount and the encoding cost for each of the plurality of search modes;
According to the selected search mode, the search range is set on the other picture, the pixel data included in the search range on the other picture is read from the storage unit, and the code is read within the search range. A motion vector calculation unit that obtains a region that most closely matches the encoding target block and calculates a vector representing a movement amount between the encoding target block and the region as the motion vector;
A moving picture encoding apparatus having: The encoding target block can be divided into a plurality of sub-blocks having different sizes,
The encoding cost calculation unit, for each of the plurality of search modes, for each combination of the subblocks, the encoding cost representing the motion vector candidate of each subblock included in the combination and the code amount of the subblock. Is determined as the encoding cost of the encoding target block, and the combination of the subblocks corresponding to the minimum value is a combination of subblocks for the search mode. age,
The memory access amount calculation unit, for each of the plurality of search modes, based on the positional relationship of the search range indicated by the motion vector candidate of each subblock included in the combination of the subblocks for the search mode Obtaining the memory access amount for the search mode;
The moving image encoding apparatus according to claim 1, wherein the motion vector calculation unit calculates the motion vector for each sub-block included in the combination of sub-blocks for the selected search mode.   The search mode determination unit selects a search mode in which the coding cost is minimum among search modes in which the memory access amount is equal to or less than a target access amount among the plurality of search modes. The moving image encoding apparatus described in 1.   The moving image encoding according to claim 3, further comprising a target access amount determination unit that increases the target access amount as the sum of the memory access amounts of the predetermined number of blocks nearest to the encoding target block decreases. apparatus.   The search mode determination unit, for any search mode in which the memory access amount is equal to or less than the target access amount among the plurality of search modes, the coding cost of the search mode and the block immediately before the block. 5. The search mode according to claim 3, wherein, when a difference from the encoding cost is out of an allowable range, a search mode that minimizes the memory access amount is selected from search modes in which the difference is included in the allowable range. Video encoding device. A video encoding method for encoding an error between a block to be encoded on a picture included in video data and a reference block indicated by a motion vector on another encoded picture,
For each of a plurality of search modes defining the motion vector search method, it corresponds to the search range of the motion vector defined in the search mode on the reduced picture obtained by reducing the number of pixels included in the other picture. A motion vector candidate for obtaining an area that most closely matches a reduced block obtained by reducing the number of pixels included in the encoding target block within the area, and specifying the search range from a movement amount between the reduced block and the most coincident area And an encoding cost representing the code amount of the encoding target block,
For each of the plurality of search modes, based on the search range in the search mode, a memory access amount representing a data amount to be read from the storage unit storing the other picture to calculate the motion vector Calculate
Based on the memory access amount and the encoding cost for each of the plurality of search modes, select a search mode to be applied to the encoding target block from the plurality of search modes,
According to the selected search mode, the search range is set on the other picture, the pixel data included in the search range on the other picture is read from the storage unit, and the code is read within the search range. A region that most closely matches the encoding target block is calculated, and a vector representing a movement amount between the encoding target block and the region is calculated as the motion vector.
A moving picture encoding method including the above. A moving picture coding computer program for coding an error between a coding target block on a picture included in moving picture data and a reference block indicated by a motion vector on another coded picture. And
For each of a plurality of search modes defining the motion vector search method, it corresponds to the search range of the motion vector defined in the search mode on the reduced picture obtained by reducing the number of pixels included in the other picture. A motion vector candidate for obtaining an area that most closely matches a reduced block obtained by reducing the number of pixels included in the encoding target block within the area, and specifying the search range from a movement amount between the reduced block and the most coincident area And an encoding cost representing the code amount of the encoding target block,
For each of the plurality of search modes, based on the search range in the search mode, a memory access amount representing a data amount to be read from the storage unit storing the other picture to calculate the motion vector Calculate
Based on the memory access amount and the encoding cost for each of the plurality of search modes, select a search mode to be applied to the encoding target block from the plurality of search modes,
According to the selected search mode, the search range is set on the other picture, the pixel data included in the search range on the other picture is read from the storage unit, and the code is read within the search range. A region that most closely matches the encoding target block is calculated, and a vector representing a movement amount between the encoding target block and the region is calculated as the motion vector.
A computer program for encoding a moving image for causing a computer to execute the above.

Images