JP2005012439A - Encoding apparatus, encoding method, and encoding program - Google Patents

Abstract

An encoding apparatus, method, and program capable of always compressing and encoding a moving image with optimal encoding efficiency without increasing the amount of information in compression encoding.
An encoding apparatus 1 that performs motion compensation prediction in units of blocks of each original image included in a moving image and compresses and encodes the moving image, and includes a macroblock dividing unit 3 that divides the moving image into small blocks. A block group integration unit 5 that integrates the block group, a block group motion vector search unit 9 that searches for a motion vector of the block group based on a reference image, and a small block motion compensation unit 17 that outputs residual information, , A motion vector encoding unit 13 to be an encoded motion vector, and a macroblock integrated information amount comparison unit 19 that selects a block group having a minimum total information amount as an optimal block group and outputs a motion vector. .
[Selection] Figure 1

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to moving image processing, and more particularly to an encoding device, an encoding method, and an encoding program that divide an original image included in a moving image into blocks and perform compression encoding using motion compensated prediction.
[0002]
[Prior art]
Conventionally, compression encoding methods for moving images are MPEG-1, 2, 4 (ISO / IEC 11172-2 and ISO / IEC 13818-2 respectively) standardized by ISO / IEC JTC1 / SC29 / WG11 (MPEG). , ISO / IEC 14496-2) and H.264 recommended by ITU-T. And JVT (Joint Video Team (MPEG-4 part 10 AVC [ISO / IEC 14496-10] / ITU-T H.264)). This compression coding processing of a moving image involves dividing an original image (image) included in the moving image into predetermined regions called blocks, and performing motion compensation prediction and DCT conversion processing in units of the divided blocks. (For example, see Non-Patent Documents 1 and 2).
[0003]
In the case of motion compensation prediction, the block size as a unit of compression encoding processing is 16 horizontal pixels × vertical 16 lines called macroblocks in the MPEG-1 and MPEG-2 systems. One motion vector composed of two components in the horizontal direction and the vertical direction is assigned to the one macroblock.
[0004]
Here, each compression encoding process will be described with reference to a schematic block diagram shown in FIG.
FIG. 4A schematically shows a motion vector assignment process in the case of motion compensation prediction in units of macroblocks as in MPEG-1 and MPEG-2. As described above, in the motion vector allocation processing in the case of motion compensation prediction, a macroblock is used as a unit of processing, and all the pixels in the macroblock are represented by the same motion vector. Then, a moving image is processed assuming a motion model in which the pixels in the macroblock are also translated in accordance with the direction of the motion vector and the amount of movement.
[0005]
Also, MPEG-4 and ITU-T H.264. In the H.263 system, in addition to performing motion compensation prediction in units of macroblocks, as shown in FIG. 4B, a mode in which motion compensation is performed in units of blocks (small blocks) each having a size of horizontal 8 pixels × vertical 8 lines ( 8 × 8 mode). When this 8 × 8 mode is used, when there are a plurality of motions in a macroblock of 16 horizontal pixels × 16 vertical lines, the actual motion is more than the case where one motion vector is assigned to one macroblock. Can be obtained.
[0006]
In other words, in the case of the 8 × 8 mode, by setting a flag indicating whether the block unit for performing motion compensation prediction is a macro book or a small block (outputs a discrimination signal), It is configured so that it can be determined at the time of decoding processing for decoding. However, when dividing into small blocks, there are four small blocks for one macroblock, so the number of motion vectors is one for each small block. There will be one.
[0007]
In addition, in the JVT method, as shown in FIGS. 4C and 4D, in addition to a square block having a size of 16 macro blocks × 16 vertical lines for a macro block and 8 horizontal pixels × 8 vertical lines for a small block. In addition, a rectangular block such as horizontal 16 pixels × vertical 8 lines and horizontal 8 pixels × vertical 16 lines, or a macroblock having a size of horizontal 8 pixels × vertical 8 lines as shown in FIGS. The blocks divided into four, and the block having the size of horizontal 8 pixels × vertical 8 lines are divided into horizontal 8 pixels × vertical 4 lines, horizontal 4 as shown in FIGS. It is also possible to perform motion compensation prediction with a block divided into a size of pixel × vertical 8 lines and horizontal 4 pixels × vertical 4 lines.
[0008]
In the case of the JTV system, a large number of flags indicating what size block the macroblock of 16 horizontal pixels × 16 vertical lines is to be divided are required.
[0009]
For example, when a macro block of 16 horizontal pixels × vertical 16 lines is divided into two rectangular blocks of horizontal 16 pixels × vertical 8 lines, two motion vectors exist in one macro block. Thus, the number of motion vectors is smaller than when the macroblock is divided into four square blocks.
[0010]
By the way, in each of these compression encoding systems, as shown in FIG. 5, the motion vector MV (X) estimated for each macroblock or each small block obtained by dividing the macroblock is shown in one block. The motion vectors MV (A), MV (B), MV (C), and MV (D (D) in adjacent blocks A, B, C, and D are not encoded within the processing target block X in FIG. ) Is approximated, a motion vector difference (dMV (X)) is taken between adjacent blocks, and encoding is often performed using this difference (dMV (X)). For this reason, the amount of information (encoded information amount) generated after encoding becomes smaller as the direction and size of the motion vector are approximated between adjacent blocks.
[0011]
In FIG. 5, a portion surrounded by a broken line indicates one frame. Further, the processing target block X is located at an arbitrary position in this frame, and although not shown in FIG. 5, it is usually necessary to use all the blocks surrounding the processing target block X (adjacent blocks). There is.
[0012]
[Non-Patent Document 1]
Atsushi Kato et al. “Motion compensation based on block unit boundary model: Refinement of H.26L prediction method” Proceedings of the IEICE Conference, March 29, 2002, pp401-pp402
[Non-Patent Document 2]
Seiji Kimura et al. “Motion Compensation Method Based on Variable Size and Variable Shape Block” IEICE Transactions, February 1997 pp434-pp443
[0013]
[Problems to be solved by the invention]
However, the conventional compression coding method that performs motion compensation prediction in units of blocks of each original image included in a moving image has the following problems.
[0014]
MPEG-4, ITU-TH, which performs motion compensation prediction by dividing a macro block into small blocks. Although the H.263 and JVT systems are more likely to be able to compress and encode a moving image with a motion vector that matches the actual motion than the MPEG-1 and 2 systems, which can perform motion compensation prediction only in units of macroblocks. Therefore, the motion vector for each divided small block and additional information indicating how the macroblock is divided are required. Therefore, the amount of information is slightly increased as compared with the MPEG-1 and 2 systems in which one motion vector is assigned to one macroblock and compression coding is performed, thereby improving coding efficiency. However, there is a problem that it is not always efficient.
[0015]
Therefore, the object of the present invention is to solve the above-mentioned problems of the prior art, and to always compress and encode a moving image with optimum encoding efficiency without increasing the amount of information at the time of compression encoding. An encoding device, an encoding method, and an encoding program are provided.
[0016]
[Means for Solving the Problems]
In order to achieve the above-described object, the present invention has the following configuration.
The encoding apparatus according to claim 1 is an encoding apparatus that performs motion compensation prediction in units of blocks of each original image included in a moving image and compresses and encodes the moving image, and includes a macroblock dividing unit and The block group integration unit, the block group motion vector search unit, the small block motion compensation unit, the motion vector encoding unit, and the macroblock integrated information amount comparison unit are provided.
[0017]
According to such a configuration, the encoding apparatus divides the macro block into small blocks by the macro block dividing unit, and integrates the block into a block group in which the small blocks are combined by the block group integrating unit. That is, a macroblock divided into 16 horizontal pixels × vertical 16 lines is divided into blocks smaller than this size, for example, four small blocks of 8 horizontal pixels × 8 vertical lines, and the macroblock is divided into four. Alternatively, the macroblock is divided into three, one small block of horizontal 8 pixels × vertical 16 lines and two small blocks of horizontal 8 pixels × vertical 8 lines. Alternatively, the macro block is divided into two by dividing into one small block of horizontal 8 pixels × vertical 8 lines and the other remaining blocks.
[0018]
Subsequently, in this encoding device, the block group motion vector search means searches for the motion vector of the block group integrated by the block group integration means based on the reference image used when performing motion compensation prediction. This search for a motion vector indicates a corresponding block group in the previous (immediately preceding) image (frame) that is a reference image. That is, in this block group motion vector search means, the motion vector of the block group which is various combinations of small blocks is searched for the number of various combinations of small blocks by referring to the corresponding block group of the immediately preceding frame. Will be.
[0019]
The encoding apparatus performs motion compensation prediction for each small block by the small block motion compensation unit using the motion vector searched by the block group motion vector search unit, and the motion compensation prediction result and the original result. Outputs residual information generated by comparison with an image. That is, this motion compensation prediction (motion compensation process) is performed for each small block constituting the block group searched by the block group motion vector search means.
[0020]
Further, the encoding apparatus encodes the motion vector searched by the block group motion vector search means by the motion vector encoding means, and outputs the residual output from the small block motion compensation means by the macroblock integrated information amount comparison means. A block in which the total amount of information of difference information is summed for each macroblock unit, and the total amount of information obtained by adding the information amount of encoded motion vectors encoded by the motion vector encoding means is the smallest among the combinations of block groups. A group is selected as the optimum block group, and a motion vector of the optimum block group is output. That is, by selecting a motion vector that provides the best coding efficiency from among a plurality of integrated block groups, it is possible to improve the coding efficiency of moving images based on the motion vector.
[0021]
The encoding method according to claim 2 is an encoding method in which motion compensation prediction is performed in units of blocks of each original image included in a moving image, and the moving image is compressed and encoded. The procedure includes a block group integration step, a block group motion vector search step, a small block motion compensation step, a motion vector encoding step, and a macroblock integration information amount comparison step.
[0022]
According to such a procedure, the encoding method divides a macroblock into small blocks in the macroblock division step, and integrates the block into a block group in which the small blocks are combined in the block group integration step. Subsequently, in the block group motion vector search step, this encoding method searches for the motion vector of the block group integrated in the block group integration step based on the reference image used when performing motion compensation prediction. In the small block motion compensation step, the encoding method performs motion compensation prediction for each small block using the motion vector searched in the block group motion vector search step, and the motion compensation prediction result Residual information generated by comparison with the original image is output.
[0023]
Also, this encoding method encodes the motion vector searched in the block group motion vector search step in the motion vector encoding step, and outputs it in the small block motion compensation step in the macroblock integrated information amount comparison step. The total amount of residual information is summed for each macroblock unit, and the total amount of information obtained by adding the encoded motion vector information amount encoded in the motion vector encoding step is the smallest among the block group combinations. Is selected as the optimum block group, and the motion vector of this optimum block group is output.
[0024]
4. The encoding program according to claim 3, wherein an apparatus for performing motion compensation prediction in units of blocks of each original image included in a moving image and compressing and encoding the moving image includes a macroblock dividing unit and a block group integrating unit. , Block group motion vector search means, small block motion compensation means, motion vector encoding means, and macroblock integrated information amount comparison means.
[0025]
According to such a configuration, the encoding program divides the macro block into small blocks by the macro block dividing unit and integrates the block into a block group in which the small blocks are combined by the block group integrating unit. Subsequently, in the encoding program, the block group motion vector search unit searches for the motion vector of the block group integrated by the block group integration unit based on the reference image used when motion compensation prediction is performed. Then, the encoding program performs motion compensation prediction for each small block using the motion vector searched by the block group motion vector search means in the small block motion compensation means, and the result of the motion compensation prediction Residual information generated by comparison with the original image is output.
[0026]
In addition, this encoding program encodes the motion vector searched by the block group motion vector search means by the motion vector encoding means, and is output by the small block motion compensation means by the macroblock integrated information amount comparison means. The total amount of information of the residual information is summed for each macroblock unit, and the total amount of information obtained by adding the information amount of the encoded motion vector encoded by the motion vector encoding means is the smallest among the combinations of the block groups. Is selected as the optimum block group, and the motion vector of this optimum block group is output.
[0027]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.
(Configuration of encoding device)
FIG. 1 is a block diagram of an encoding apparatus. As shown in FIG. 1, the encoding apparatus 1 receives a macroblock (macroblock signal) obtained by dividing each image constituting a moving image, outputs a motion vector of the macroblock, An image is encoded, and includes a macroblock division unit 3, a block group integration unit 5, a block group accumulation sequential output unit 7, a block group motion vector search unit 9, a reference image accumulation unit 11, and a motion vector code. A conversion unit 13, a motion vector storage unit 15, a small block motion compensation unit 17, and a macroblock total information amount comparison unit 19.
[0028]
The macroblock dividing unit 3 (macroblock dividing means) converts the inputted macroblock signal (horizontal 16 pixels × vertical 16 lines) into small blocks (horizontal 16 pixels × vertical 8 lines, horizontal 8 pixels × vertical 16 lines, horizontal 8). A divided block signal divided into pixels × vertical 8 lines) is output to the block group integration unit 5.
[0029]
That is, the macroblock dividing unit 3 divides the macroblock into rectangular or square small blocks in a format according to the MPEG-4 or JVT system. According to MPEG-4, a motion vector is added to each of a mode in which one macro block is expressed with one motion vector and a small block obtained by dividing the macro block into four. Mode.
[0030]
Further, in accordance with the JVT method, a macro block that does not divide horizontal 16 pixels × vertical 16 lines, a square block of horizontal 8 pixels × vertical 8 lines, horizontal 16 pixels × vertical 8 lines, and horizontal 8 pixels × vertical. A 16-line rectangular block is divided into blocks of horizontal 8 pixels × vertical 4 lines obtained by further dividing horizontal 8 pixels × vertical 8 lines, horizontal 4 pixels × vertical 8 lines, and horizontal 4 pixels × vertical 4 lines.
[0031]
The block group integration unit 5 (block group integration means) integrates the divided block signals (small blocks) obtained by dividing the macro block by the macro block division unit 3 to generate various combinations of block group signals (block groups). And output to the block group accumulation sequential output unit 7. In the block group integration unit 5, as shown in FIG. 3, a block group is generated by a combination of all small blocks.
[0032]
As shown in FIG. 3 (1), the macroblock is divided into four small blocks, and the directions and sizes of the three motion vectors are approximate, but are slightly different, so encoding is performed. A differential signal is generated.
[0033]
FIG. 3B shows a case where the macroblock is not divided (non-divided). Note that the symbols “a”, “b”, and “c” shown in FIG. 3 indicate divided areas. FIGS. 3 (3) to 3 (9) show block groups when the macroblock is divided into two, and FIGS. 3 (10) to 3 (19) show blocks when the macroblock is divided into three. Shows the group.
[0034]
In FIG. 3 (3), the upper left horizontal 8 pixels × vertical 8 lines small block is divided into areas other than this block, and conversely, in FIG. 3 (4), the upper right horizontal 8 pixels. X Divided into small 8-line vertical blocks and areas other than these blocks. Further, in FIG. 3 (5), it is divided into a small block of horizontal 8 pixels × vertical 8 lines in the lower left and an area other than this block. Conversely, in FIG. It is divided into a small block of 8 pixels × vertical 8 lines and an area other than this block.
[0035]
In FIG. 3 (7), it is divided into two small blocks of horizontal 16 pixels × vertical 8 lines, and in FIG. 3 (8), it is divided into two small blocks of horizontal 8 pixels × vertical 16 lines. In FIG. 3 (9), it is divided into 4 blocks of 8 blocks horizontal x 8 vertical blocks. However, since the divided areas are “a” and “b” areas, they are divided into two as a result. Will be.
[0036]
In FIG. 3 (10), the left half is divided into small blocks of 16 horizontal pixels × vertical 8 lines, and the right half is divided into 2 small blocks of 8 horizontal pixels × vertical 8 lines. In FIG. 3 (11), the left half is divided into two small blocks of horizontal 8 pixels × vertical 8 lines, and the right half is divided into small blocks of horizontal 16 pixels × vertical 8 lines.
[0037]
In FIG. 3 (12), the upper half is divided into small blocks of 8 horizontal pixels × vertical 16 lines, and the lower half is divided into 2 small blocks of 8 horizontal pixels × vertical 8 lines. In FIG. 3 (13), the upper half is divided into two small blocks of 8 horizontal pixels × vertical 8 lines, and the lower half is divided into small blocks of 8 horizontal pixels × vertical 16 lines.
[0038]
3 (14) and 16 (16), it is divided into four blocks of horizontal 8 pixels × vertical 8 lines, and the upper right part and the lower left part are the same area. As a result, “a”, “b”, “ It is divided into three areas c ″. 3 (15) and 3 (17), the block is divided into four blocks of horizontal 8 pixels × vertical 8 lines, and the upper left and lower right are the same area. As a result, “a”, “b” , “C” area is divided into three.
[0039]
Returning to FIG. 1, the description will be continued.
The block group accumulation sequential output unit 7 accumulates the block group signal (block group) integrated by the block group integration unit 5 and sequentially stores the accumulated block group signal (block group) in the block group motion vector search unit 9. Output. For example, the block group is sequentially (in order) output from the block group accumulation sequential output unit 7 immediately after the block group motion vector input to the block group motion vector search unit 9 is searched.
[0040]
The block group motion vector search unit 9 (block group motion vector search means) receives the motion vector for each small block constituting the input block group as a reference image signal (reference image, one image) stored in the reference image storage unit 11. The previous frame) is searched and output to the motion vector encoding unit 13 and the motion vector storage unit 15 as a motion vector signal. In this embodiment, the block group motion vector search unit 9 searches for a block group motion vector by the block matching method.
[0041]
The reference image accumulating unit 11 accumulates, as a reference image, an image of a frame immediately before (immediately before) a macroblock signal obtained by dividing each image included in the moving image input to the apparatus 1. The reference image storage unit 11 is usually composed of a frame memory.
[0042]
The motion vector encoding unit 13 outputs an encoded motion vector obtained by encoding the motion vector signal (motion vector for each small block) output from the block group motion vector search unit 9 to the macroblock integrated information amount comparison unit 19. Is. The motion vector encoding unit 13 takes a difference between the motion vector of the small block constituting the block group input to the block group motion vector search unit 9 and the motion vector of the macro block adjacent to the macro block, The difference is encoded.
[0043]
The motion vector accumulation unit 15 accumulates the motion vector of the block group searched by the block group motion vector search unit 9.
[0044]
The small block motion compensation unit 17 is a block group motion vector searched by the block group motion vector search unit 9 and stored in the motion vector storage unit 15, and a reference image signal output (supplied) from the reference image storage unit 11. (Reference image) is used to perform motion compensation processing (output of motion compensation residual signal [residual information]) for each small block.
[0045]
The macroblock total information amount comparison unit 19 includes a motion compensation residual signal (residual information) output from the small block motion compensation unit 17, a macroblock signal (macroblock) input to the apparatus 1, and a motion vector. Based on the encoded motion vector signal (encoded motion vector) output from the encoding unit 13, the block group (see FIG. 3) integrated by the block group integration unit 5 has the highest encoding efficiency. And a motion vector obtained in this block group is output as a final motion vector signal.
[0046]
That is, in the macroblock integrated information amount comparison unit 19, first, using the result of motion compensation for each small block in the small block motion compensation unit 17, the reference image, that is, the immediately preceding original image included in the moving image is used. The information amount of the residual information (motion compensation residual signal) generated by the comparison is totaled in the macroblock. Next, a total information amount obtained by adding the information amount of the encoded motion vector (encoded motion vector signal) encoded by the motion vector encoding unit 13 to the total information amount is obtained. Then, a block group is selected that is a combination of small blocks that minimizes the total information amount (minimum encoded information amount). That is, the block group integration unit 5 selects an optimal combination from the block groups in which the small blocks are combined, and the motion vector of the block group obtained by this combination is finally output (final motion vector signal). It becomes.
[0047]
According to this encoding apparatus 1, the macroblock dividing unit 3 divides the macroblock into small blocks, and the block group integrating unit 5 integrates the block into a block group obtained by combining the small blocks. Subsequently, the block group motion vector search unit 9 searches the block group motion vector integrated by the block group integration unit 5 and stored by the block group storage sequential output unit 7 based on the reference image. Then, the motion compensation prediction is performed for each small block by the motion vector searched by the block group motion vector search unit 9 by the small block motion compensation unit 17 and the result of the motion compensation prediction is compared with the reference image. The generated residual information is output.
[0048]
Further, the motion vector searched by the block group motion vector search unit 9 is encoded by the motion vector encoding unit 13, and the residual output from the small block motion compensation unit 17 by the macroblock integrated information amount comparison unit 19 is encoded. A block in which the total amount of information is summed for each macroblock unit, and the total amount of information obtained by adding the information amount of the encoded motion vector encoded by the motion vector encoding unit 13 is the smallest among the combinations of the block groups. A group is selected as the optimum block group, and a motion vector of the optimum block group is output. That is, by selecting a motion vector having the best encoding efficiency from among the block groups integrated in a plurality of ways by the block group integration unit 5, the encoding efficiency of the moving image is improved based on this motion vector. be able to. That is, it is possible to always compress and encode a moving image with optimum encoding efficiency without increasing the amount of information when the moving image is encoded (compression encoded).
[0049]
(Operation of encoding device)
Next, the operation of the encoding apparatus 1 will be described with reference to the flowchart shown in FIG. 2 (see FIG. 1 as appropriate).
[0050]
First, the macroblock signal (macroblock) inputted by the macroblock dividing unit 3 in accordance with the encoding method (MPEG-4, JVT method) is horizontal 8 pixels × vertical 8 lines, horizontal 16 pixels × vertical. Divided into small blocks of 8 lines, horizontal 8 pixels × vertical 16 lines, etc. (S1). Subsequently, the block group integration unit 5 integrates the small blocks divided by the macroblock division unit 3 into various combined block groups (generates a block group) (S2).
[0051]
Then, the block group motion vector search unit 9 searches for a motion vector for each block group (S3). Then, the small block motion compensation unit 17 performs a motion compensation process (motion compensation prediction) for each small block, and outputs residual information (compensation error) to the macroblock integrated information amount comparison unit 19 (S4). Also, the motion vector encoding unit 13 outputs the encoded motion vector obtained by encoding the motion vector to the macroblock integrated information amount comparison unit 19.
[0052]
Then, the macroblock total information amount comparison unit 19 adds the information amount of residual information (compensation error) in the macroblock, and the total information amount obtained by adding the information amount of the encoded motion vector to the total information amount. Are compared with the information amount of the block group at the corresponding portion of the reference image, and the smaller information amount is selected and held (S5).
[0053]
The process of S5 is performed for all block groups (all combinations) integrated by the block group integration unit 5, and it is determined whether or not all combinations have been completed (S6), and it is determined that all combinations have been completed. If not (S6, No), the process returns to S2, and if it is determined that all combinations have been completed (S6, Yes), the block information group whose total information amount (encoded information amount) is minimized A motion vector is output (S7). For example, as shown in FIG. 3A, when there are four motion vectors, that is, when the three motion vectors are directed in the upper right direction and one motion vector is directed in the left direction. In addition, the final motion vector is obtained by shortening the length of the motion vector shown in the upper right.
[0054]
As mentioned above, although this invention was demonstrated based on one Embodiment, this invention is not limited to this.
For example, the processing of each component of the encoding device 1 can be regarded as an encoding method that considers each process as one process, or the encoding of the processing of each component of the encoding device 1 described in a general-purpose computer language It can also be regarded as a program. In these cases, the same effect as that of the encoding device 1 can be obtained.
[0055]
【The invention's effect】
According to the first, second, and third aspects of the present invention, the motion vector having the highest coding efficiency is selected from among a group of blocks obtained by integrating a plurality of small blocks obtained by dividing a macroblock. Therefore, it is possible to improve the coding efficiency of moving images, and always compress and code moving images with optimal coding efficiency without increasing the amount of information when coding (compressing) moving images. Can be
[Brief description of the drawings]
FIG. 1 is a block diagram of an encoding apparatus according to an embodiment of the present invention.
FIG. 2 is a flowchart for explaining the operation of the encoding apparatus shown in FIG. 1;
FIG. 3 is a diagram illustrating a block group in which small blocks are integrated by a block group integration unit.
FIG. 4 is a diagram illustrating division of a conventional macroblock.
FIG. 5 is a diagram illustrating encoding of a conventional motion vector.
[Explanation of symbols]
1 Encoder
3 Macroblock division (macroblock division means)
5 Block group integration unit (block group integration means)
7 Block group accumulation sequential output section
9 Block group motion vector search unit (block group motion vector search means)
11 Reference image storage unit
13 Motion vector encoding unit (motion vector encoding means)
15 Motion vector storage unit
17 Small block motion compensation unit (small block motion compensation means)
19 Macroblock integrated information amount comparison unit (macroblock integrated information amount comparison means)

Claims (3)

An encoding device that performs motion compensation prediction in units of blocks of each original image included in a moving image and compresses and encodes the moving image,
Macroblock dividing means for dividing the macroblock obtained by dividing the original image into small blocks that are smaller than the macroblock;
A block group integrating means for integrating the block blocks combined by the macroblock dividing means into a combined block group;
A block group motion vector search means for searching for a motion vector of the block group integrated by the block group integration means based on a reference image used in the motion compensation prediction;
The motion vector searched by the block group motion vector search means performs motion compensation prediction for each small block, and outputs a residual information generated by comparing the result of the motion compensation prediction with the original image. Block motion compensation means;
A motion vector encoding unit that uses the motion vector searched by the block group motion vector search unit as an encoded motion vector;
The total amount of information of the residual information output by the small block motion compensation unit is totaled for each macroblock, and the total amount of information obtained by adding the information amount of the encoded motion vector encoded by the motion vector encoding unit is Selecting a block group that is the smallest among the combinations of the block groups as an optimum block group, and outputting a motion vector of the optimum block group, a macroblock integrated information amount comparison unit;
An encoding device comprising: An encoding method that performs motion compensation prediction in units of blocks of each original image included in a moving image and compresses and encodes the moving image,
A macroblock dividing step of dividing the macroblock obtained by dividing the original image into small blocks that are smaller than the macroblock;
A block group integration step of integrating the small blocks divided in this macroblock division step into a block group that is combined;
A block group motion vector search step for searching for a motion vector of the block group integrated in the block group integration step based on a reference image used in the motion compensation prediction;
The motion vector searched in the block group motion vector search step is subjected to motion compensation prediction for each small block, and the residual information generated by comparing the result of the motion compensation prediction with the original image is output. A block motion compensation step;
A motion vector encoding step in which the motion vector searched in the block group motion vector search step is an encoded motion vector encoded;
The total amount of information of the residual information output in the small block motion compensation step is totaled for each macroblock, and the total amount of information obtained by adding the information amount of the encoded motion vector encoded in the motion vector encoding step is Selecting a block group that is the smallest among the combinations of the block groups as an optimal block group, and outputting a motion vector of the optimal block group, a macroblock integrated information amount comparison step;
The encoding method characterized by including. A device that performs motion compensation prediction in units of blocks of each original image included in the moving image and compresses and encodes the moving image,
Macroblock dividing means for dividing the macroblock obtained by dividing the original image into small blocks that are smaller than the macroblock;
A block group integration unit that integrates a small block divided by the macroblock division unit into a combined block group;
A block group motion vector search means for searching for a motion vector of the block group integrated by the block group integration means based on a reference image used in the motion compensation prediction;
The motion vector searched by the block group motion vector search means performs motion compensation prediction for each small block, and outputs a residual information generated by comparing the result of the motion compensation prediction with the original image. Block motion compensation means,
A motion vector encoding means for encoding a motion vector searched by the block group motion vector search means;
The total amount of information of the residual information output by the small block motion compensation unit is totaled for each macroblock, and the total amount of information obtained by adding the information amount of the encoded motion vector encoded by the motion vector encoding unit is Selecting a block group that is the smallest among the combinations of the block groups as an optimum block group, and outputting a motion vector of the optimum block group, a macroblock integrated information amount comparison unit;
An encoding program characterized in that it is made to function as:

Images