WO2011064926A1

WO2011064926A1 - Image coding device, image decoding device, image coding method, and image decoding method

Info

Publication number: WO2011064926A1
Application number: PCT/JP2010/005547
Authority: WO
Inventors: 憲道日和佐; 山田　悦久; 関口　俊一; 杉本　和夫; 彰峯澤
Original assignee: 三菱電機株式会社
Priority date: 2009-11-30
Filing date: 2010-09-10
Publication date: 2011-06-03

Abstract

A quantization control section (33) refers to a transform table which is related to a sub-block used for motion-compensated prediction processing conducted in creating a predicted image and which is one of a plurality of transform tables stored in a transform table storage section (32), and transforms a quantization parameter corresponding to a macroblock into a quantization parameter corresponding to the sub-block. Using the quantization parameter corresponding to the sub-block, a quantization section (18) quantizes an orthogonal coefficient of a differential image output from an orthogonal transform section (17).

Description

Image encoding device, image decoding device, image encoding method, and image decoding method

The present invention relates to an image encoding device and an image encoding method for encoding an image, and an image decoding device and an image decoding method for decoding an image encoded by the image encoding device.

For example, FIG. 2 is an explanatory diagram showing a state of quantization processing described in a H.264 (MPEG-4AVC) recommendation (ITU-T Rec. H.264).
H. In H.264, a frame or field to be encoded is divided into blocks called “macroblocks” composed of pixels of 16 pixels × 16 lines, and a quantization parameter QP that can be changed for each macroblock is used. Quantize the orthogonally transformed coefficients.
However, when quantization is performed on a 4-pixel × 4-line orthogonal transform block or an 8-pixel × 8-line orthogonal transform block (see FIG. 26), all the blocks are used. Quantization is performed using a common quantization parameter QP.

FIG. 2 is an explanatory diagram illustrating a partition of a macroblock used when performing motion compensation prediction processing defined in H.264.
H. In H.264, as shown in FIG. 2, four types of block configurations are defined as macroblock partitions.
That is, one block composed of 16 pixels × 16 lines, two blocks composed of 16 pixels × 8 lines, two blocks composed of 8 pixels × 16 lines, and four blocks composed of 8 pixels × 8 lines. Configuration is specified.

Further, in the case of a block composed of 8 pixels × 8 lines, three types of sub-macroblock partitions are defined as shown in FIG.
That is, the configuration of two blocks of 8 pixels × 4 lines, two blocks of 4 pixels × 8 lines, and four blocks of 4 pixels × 4 lines is defined.
The number in the partition of the sub macroblock is a block index for specifying the position of the partition.
Even if the sub-macroblock partitions are different, the quantization for orthogonal transform blocks in the same macroblock is always performed using the same quantization parameter.

When there is a complicated pattern or a flat pattern in a part of a macroblock, the subjective image quality of the part can be improved if the quantization parameter of the block including each pattern can be changed.
However, if the quantization parameter is changed for each subblock (orthogonal transform block, macroblock partition, or submacroblock partition) in the macroblock, the amount of information for the quantization parameter to be transmitted increases.

Patent Document 1 below discloses a method for changing a quantization parameter for an orthogonal transform block.
FIG. 27 is a table showing variation values from quantization parameters for four orthogonal transform blocks of 8 × 8 pixels indicated by block numbers in a macroblock.
In this table, the quantized code index determined to be optimal is transmitted for each slice (area composed of one or more continuous macroblocks defined in H.264) or for each macroblock. Is done.

FIG. 28 is an explanatory diagram showing a quantization method for the orthogonal transform block.
In FIG. 28, QP is a quantization parameter transmitted for each macroblock, and V _blk = {1, 1, −1, −1} is a combination of quantization parameter fluctuation values specified by the quantization code index. Show.
That is, each of the 8 × 8 pixel orthogonal transform blocks is quantized with quantization parameters of QP + 1, QP + 1, QP-1, and QP-1.
The 4 × 4 pixel orthogonal transform block indicates that the quantization is performed using the quantization parameter of the 8 × 8 pixel orthogonal transform block including the 4 × 4 pixel orthogonal transform block.

FIG. 29 is a block diagram showing a conventional image encoding device, which realizes the quantization method described above.
The outline of the operation of this image encoding apparatus will be described below.
When the prediction signal generation unit 201 inputs an input image to be encoded in units of macroblocks, the prediction signal generation unit 201 receives the H.264 signal. A prediction image is generated in all prediction modes defined in H.264, and a difference image between the prediction image and the input image is output to the orthogonal transform unit 203 via the quantization set selection unit 202.

Upon receiving the difference image between the predicted image and the input image, the orthogonal transform unit 203 performs orthogonal transform on the difference image and outputs the transform coefficient of the difference image to the quantization unit 204.
When the quantization unit 204 receives the transform coefficient of the difference image from the orthogonal transform unit 203, the quantization unit 204 refers to the quantization code table set by the encoding control unit 206 and performs the orthogonal transform when the orthogonal transform unit 203 performs the orthogonal transform. A quantization parameter corresponding to the orthogonal transform block to be used is acquired, and the transform coefficient of the difference image is quantized in units of orthogonal transform blocks using the quantization parameter.
When variable length coding section 205 receives the quantized transform coefficient from quantization section 204, variable length coding section 205 performs variable length coding on the transform coefficient.

As a result, the transform coefficient of the difference image is quantized in units of orthogonal transform blocks, but is quantized in units of sub-macroblocks used when performing motion compensation prediction processing that significantly reflects the design. It is not a thing.
In a conventional image coding apparatus, a quantization code table must be created by always using four 8 × 8 pixel orthogonal transform blocks in a macroblock as a set, and the number of combinations of quantization parameter variation values is increased. As a result, the number of indexes in the quantization code table becomes enormous.
Conversely, if the number of indexes in the quantization code table is reduced, the number of combinations of variation values of quantization parameters is extremely reduced.

H. In conventional international video standards such as H.264, the block unit to be intra-coded is the same as that of the macroblock. However, in future standards, the intra-code is used for each block in the macroblock. It is also possible to perform the conversion.
In this case, in the same macroblock, a block to be encoded with intra and a block to be encoded with inter are mixed.
In intra coding and inter coding, due to the difference in coding efficiency, if quantization is performed using the same quantization parameter, a large difference occurs in the amount of generated code. Thus, when a block that performs intra coding and a block that performs inter coding coexist in the same macroblock, it is desirable to perform quantization with different quantization parameters.

JP 2006-262004 A (FIGS. 1, 3 and 14)

Since the conventional image coding apparatus is configured as described above, the quantization parameter cannot be changed in units of sub-macroblocks used when performing motion compensation prediction processing that significantly reflects the pattern. Also, unless the number of indexes in the quantization code table is increased enormously, the number of combinations of quantization parameter fluctuation values cannot be increased. For this reason, there is a problem that it is difficult to improve the image quality of the sub-macroblock whose image quality is visually significantly deteriorated in the macroblock.

In addition, if quantization is performed using the same quantization parameter for all the blocks in a macro block as in the case of the encoding method of the international standard, the intra block and the inter block are included in the same macro block. When the blocks are mixed, it is difficult to control the code amount, and there is a problem that the image quality may be deteriorated.

The present invention has been made to solve the above-described problems, and an image encoding device and an image encoding method capable of improving the image quality of a sub-block that is visually noticeably deteriorated in a macro block. The purpose is to obtain.
Another object of the present invention is to obtain an image decoding device and an image decoding method applicable to the image encoding device.
Also, the present invention provides an image encoding device capable of preventing image quality deterioration by performing quantization on each block with different quantization parameters when an intra block and an inter block are mixed in the same macro block. An object is to obtain an image decoding device, an image encoding method, and an image decoding method.

The image coding apparatus according to the present invention records a correspondence relationship between a quantization parameter corresponding to a macroblock and a quantization parameter corresponding to the subblock for each of various subblocks having a size smaller than that of the macroblock. A motion compensation prediction process performed when a prediction image is generated by the difference image generation unit among the conversion table storage unit storing the conversion table and the plurality of conversion tables stored by the conversion table storage unit. A quantization parameter conversion unit that converts a quantization parameter corresponding to the macroblock into a quantization parameter corresponding to the subblock with reference to a conversion table related to the subblock to be used, and the quantization unit generates a difference image The difference image generated by the means is orthogonally transformed and transformed by the quantization parameter transformation means Using a quantization parameter corresponding to the sub-blocks, the transform coefficients of the difference image is obtained so as to quantization.

According to the present invention, for each of various sub-blocks having a size smaller than that of the macro block, the conversion table that records the correspondence between the quantization parameter corresponding to the macro block and the quantization parameter corresponding to the sub-block is stored. Sub-table used in motion compensation prediction processing performed when a prediction image is generated by the difference image generation means among the plurality of conversion tables stored by the conversion table storage means and the conversion table storage means A quantization parameter conversion unit that converts a quantization parameter corresponding to the macroblock into a quantization parameter corresponding to the sub-block with reference to the conversion table related to the block, and the quantization unit is generated by the difference image generation unit; The sub-blocks obtained by orthogonal transformation of the difference image and transformed by the quantization parameter transformation means Since the conversion coefficient of the difference image is quantized using the quantization parameter corresponding to the image, the image quality of the sub-block where the image quality degradation is visually noticeable in the macro block can be improved. effective.

It is a block diagram which shows the image coding apparatus by Embodiment 1 of this invention. H. 2 is an explanatory diagram illustrating a partition of a macroblock used when performing a motion compensation prediction process defined in H.264. H. 2 is an explanatory diagram showing sub-block partitions used when performing motion compensation prediction processing defined in H.264. It is explanatory drawing which shows an example of the conversion table which concerns on the partition of block index = 0 in the macroblock (refer FIG.2 (b)) which consists of two 16x8 blocks. It is explanatory drawing which shows the example in which characters, such as a caption superimposition, are multiplexed on the input image (moving image). It is explanatory drawing which shows an example of a table ID table. It is a block diagram which shows the image decoding apparatus by Embodiment 1 of this invention. It is a flowchart which shows the processing content of the image coding apparatus by Embodiment 1 of this invention. It is a flowchart which shows the processing content of the image decoding apparatus by Embodiment 1 of this invention. It is a block diagram which shows the image coding apparatus by Embodiment 4 of this invention. It is a block diagram which shows the image decoding apparatus by Embodiment 4 of this invention. It is a block diagram which shows the image coding apparatus by Embodiment 5 of this invention. It is explanatory drawing which shows an example of the intra block in a macroblock. It is a block diagram which shows the image decoding apparatus by Embodiment 5 of this invention. It is a block diagram which shows the image coding apparatus by Embodiment 6 of this invention. H. 2 is an explanatory diagram illustrating an intra prediction type defined in H.264. It is explanatory drawing which shows table ID corresponding to an intra prediction type and a division | segmentation block. It is explanatory drawing which shows the example which performs the quantization using the quantization parameter to which the transition value QP_offset is applied with respect to the division block in the block encoded with an intra. It is explanatory drawing which shows the example which performs quantization using the quantization parameter to which the transition value QP_offset is applied with respect to the block coded by intra. It is a block diagram which shows the image decoding apparatus by Embodiment 6 of this invention. It is a block diagram which shows the image coding apparatus by Embodiment 7 of this invention. It is explanatory drawing which shows the orthogonal transformation mode of an intra block. It is explanatory drawing which shows table ID corresponding to the orthogonal transformation mode and orthogonal transformation block of an intra block. It is a block diagram which shows the image decoding apparatus by Embodiment 7 of this invention. H. It is explanatory drawing which shows the mode of the quantization process described in the H.264 recommendation. It is explanatory drawing which shows the orthogonal transformation block which performs quantization inside a macroblock. It is a table figure which shows the variation value from the quantization parameter with respect to four 8x8 orthogonal transformation blocks shown by a block number within a macroblock. It is explanatory drawing which shows the method of quantization with respect to an orthogonal transformation block. It is a block diagram which shows the conventional image coding apparatus.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
Embodiment 1 FIG.
FIG. 1 is a block diagram showing an image coding apparatus according to Embodiment 1 of the present invention.
In FIG. 1, H.264, which is an international standard for image coding systems. An example of an image encoding apparatus applied to H.264 (MPEG-4AVC) is shown, but an image encoding apparatus applied to another image encoding method may be used.
In FIG. 1, an encoding unit 1 performs an encoding process on an input image in units of macroblocks under the instruction of the encoding control unit 2.
The encoding control unit 2 performs quantization corresponding to a sub-block used when the motion compensation prediction process is performed by the encoding unit 1 and an orthogonal transform block used when a differential image described later is orthogonally transformed. Processing such as outputting parameters to the encoding unit 1 is performed.

The intra prediction unit 11 of the encoding unit 1 selects an optimal intra prediction mode for each macroblock with respect to the input image, and local decoded images (encoded codes) that have already been encoded according to the intra prediction mode. A prediction image is generated from a local decoded image of a block or macroblock in the vicinity of the conversion target block, and a process of generating a difference image between the input image and the prediction image is performed.
The motion search unit 12 performs a motion search by comparing the locally decoded image stored in the frame memory 23 with respect to the input image in units of macroblocks, and performs a process of calculating a motion vector.

The motion compensated prediction unit 13 uses the motion vector calculated by the motion search unit 12 for the locally decoded image stored in the frame memory 23 in units of subblocks that are the same as the macroblock or smaller in size than the macroblock. By performing the motion compensation prediction process, a process for generating a predicted image is performed. In addition, a process of outputting partition information indicating a sub-block used when performing the motion compensation prediction process to the quantization control unit 33 of the encoding control unit 2 is performed.
The differentiator 14 performs a process of generating a difference image by obtaining a difference between the input image and the predicted image generated by the motion compensation prediction unit 13.
The intra prediction unit 11, the motion search unit 12, the motion compensation prediction unit 13, and the differentiator 14 constitute a difference image generation unit.

The intra / inter determination unit 15 compares the prediction image generated by the intra prediction unit 11 with the prediction image generated by the motion compensation prediction unit 13 to determine an optimal prediction image, and the optimal prediction image is added to the adder 21. While outputting, the process which outputs the determination result which shows an optimal estimated image to the switch 16 and the entropy encoding part 24 is implemented.
If the switch 16 indicates that the determination result output from the intra / inter determination unit 15 indicates that the prediction image generated by the intra prediction unit 11 is an optimal prediction image, the difference generated by the intra prediction unit 11 If an image is selected and output to the orthogonal transform unit 17 and indicates that the prediction image generated by the motion compensation prediction unit 13 is an optimal prediction image, the difference image generated by the subtractor 14 is selected. The process of outputting to the orthogonal transform unit 17 is performed.

The orthogonal transform unit 17 performs orthogonal transform on the difference image output from the switch 16, outputs the transform coefficient of the difference image to the quantization unit 18, and uses the orthogonal transform block used when the difference image is orthogonally transformed. The process which outputs the orthogonal transformation block information which shows to the quantization control part 33 of the encoding control part 2 is implemented.
The quantization unit 18 uses the quantization parameter corresponding to the orthogonal transform block transformed by the quantization control unit 33 of the encoding control unit 2 to convert the transform coefficient of the difference image output from the orthogonal transform unit 17 into the orthogonal transform block. A process of quantizing in units is performed.
Note that the orthogonal transform unit 17 and the quantization unit 18 constitute quantization means.

The inverse quantization unit 19 performs a process of inversely quantizing the transform coefficient quantized by the quantization unit 18 and outputting the transform coefficient corresponding to the transform coefficient output from the orthogonal transform unit 17 to the inverse orthogonal transform unit 20. .
The inverse orthogonal transform unit 20 performs an inverse orthogonal transform on the transform coefficient output from the inverse quantization unit 19 and outputs a difference image corresponding to the difference image output from the switch 16 to the adder 21.

The adder 21 performs processing for adding the difference image output from the inverse orthogonal transform unit 20 and the prediction image selected by the intra / inter determination unit 15 to generate a local decoded image.
The deblocking filter unit 22 performs a deblocking filter process on the local decoded image generated by the adder 21, compensates for distortion due to compression, and stores the local decoded image after distortion compensation in the frame memory 23. To do.
The frame memory 23 is a recording medium that stores a locally decoded image after distortion compensation.

The entropy encoding unit 24 is a set of a transform coefficient quantized by the quantization unit 18 and a table ID and a table offset value indicating a transform table output from the quantization control unit 33 of the encoding control unit 2, a flag, and a macro The quantization parameter QP corresponding to the block, the determination result output from the intra / inter determination unit 15, and the prediction image generation information used for generating the optimal prediction image (output from the intra / inter determination unit 15) If the determination result indicates that the prediction image generated by the intra prediction unit 11 is the optimal prediction image, the intra prediction mode selected by the intra prediction unit 11 and the prediction generated by the motion compensation prediction unit 13 If the image indicates that it is an optimal predicted image, the motion vector calculated by the motion search unit 12) And copy coding carries out a process of generating a bitstream. The entropy encoding unit 24 constitutes entropy encoding means.
The transmission buffer 25 is connected to an external transmission means such as a line or a storage medium, for example, and temporarily holds the bit stream generated by the entropy encoding unit 24 and then performs processing for outputting the bit stream. To do.

The character detection unit 31 of the encoding control unit 2 detects a character included in an input image in units of frames or macroblocks, for example, specifies a sub-block around the upper end of a character that generates a lot of mosquito noise. Then, processing for outputting character detection information of a block index indicating the presence / absence of character detection, the type of sub-block including the character when the character is detected, and the position of the sub-block is performed.
The conversion table storage unit 32 stores a conversion table that records the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the subblock for each of the various subblocks having a size smaller than the macroblock. In addition, for each of various orthogonal transform blocks, a recording medium storing a transform table that records the correspondence between the quantization parameter corresponding to the macro block and the quantization parameter corresponding to the orthogonal transform block. is there. The conversion table storage unit 32 constitutes conversion table storage means.

The quantization control unit 33 determines the quantum corresponding to the macro block from the buffer amount of the bit stream accumulated by the transmission buffer 25, the target code amount for each macro block, the code amount of the bit stream actually generated, and the like. A process for determining the activation parameter is performed.
Also, the quantization control unit 33 refers to the conversion table related to the subblock indicated by the partition information output from the motion compensation prediction unit 13 among the plurality of conversion tables stored in the conversion table storage unit 32, and The quantization parameter corresponding to the block is converted into the quantization parameter corresponding to the subblock, or the macro is referred to by referring to the conversion table related to the orthogonal transform block indicated by the orthogonal transform block information output from the orthogonal transform unit 17. A process of converting the quantization parameter corresponding to the block into a quantization parameter corresponding to the orthogonal transform block is performed. In this embodiment, the quantization parameter corresponding to the macroblock is changed to the quantization parameter corresponding to the subblock with reference to the conversion table related to the subblock indicated by the partition information output from the motion compensation prediction unit 13. Although the case of transforming is shown, the obtained quantization parameter is commonly used for all orthogonal transform blocks included in the partition.

In addition, when the character detection information is output from the character detection unit 31, the quantization control unit 33 stores the type of the subblock including the character and the block index indicating the position of the subblock for one frame. To do. In the frame, the sub-block type and the block index of the sub-block are ranked in descending order of the number of detected characters. A table offset value indicating the amount of change in the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the subblock for all subblock types and the block index of the subblock according to the order. To decide. A corresponding table offset value is obtained from the determined table offset values based on the type of sub-block input from the motion compensation prediction unit 13 and its sub-block index (partition information). The correspondence relationship between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the sub-block is changed according to the change amount indicated by the table offset value, and the quantization corresponding to the macroblock is changed according to the changed correspondence relationship. The parameter is converted into a quantization parameter corresponding to the sub-block. The quantization control unit 33 constitutes a quantization parameter conversion unit.

In the example of FIG. 1, the character detection unit 31 gives a table offset value to the quantization control unit 33, but a sub-block whose image quality is deteriorated compared to other sub-blocks is specified, and A processing unit other than the character detection unit 31 may be mounted as long as it is a processing unit that gives the type of sub-block and its sub-block index to the quantization control unit 33. The table offset value is determined before the encoding of the encoding target frame or slice is started. When the character detection information is detected in advance before the encoding of the encoding target frame or slice is started in the character detection unit 31, the table offset value determined for the frame or slice is applied. When the character detection information of the encoding target frame or slice is not known in advance, the table offset value determined by the character detection information of the previous frame is used.

FIG. 7 is a block diagram showing an image decoding apparatus according to Embodiment 1 of the present invention.
In FIG. 7, the reception buffer 51 performs a process of receiving the bit stream generated by the image encoding device of FIG. 1 and outputting the bit stream to the entropy decoding unit 52.
The entropy decoding unit 52 entropy-decodes the bit stream output from the reception buffer 51, and dequantizes the transform coefficient quantized by the quantization unit 18 of the image encoding device in FIG. The quantization parameter corresponding to the table ID and table offset value set indicating the conversion table output from the quantization control unit 33, the flag, and the macroblock is output to the inverse quantization parameter generation unit 54. The process of outputting the determination result of the intra / inter determination unit 15 to the switch 59 is performed. In addition, if the determination result of the intra / inter determination unit 15 indicates that the prediction image generated by the intra prediction unit 11 is an optimal prediction image, the intra prediction mode that is the prediction image generation information is set to intra prediction. If it is output to the image generation unit 57 and the determination result of the intra / inter determination unit 15 indicates that the prediction image generated by the motion compensation prediction unit 13 is an optimal prediction image, the prediction image generation information is used. A process of outputting the type of a certain sub-block and its block index and motion vector to the inter predicted image generation unit 58 is performed. The entropy decoding unit 52 constitutes entropy decoding means.

The conversion table storage unit 53 is a recording medium that stores the same conversion table as the conversion table storage unit 32 in the image encoding device of FIG. That is, for each of the various sub-blocks having a size smaller than the macro block, a conversion table that records the correspondence between the quantization parameter corresponding to the macro block and the quantization parameter corresponding to the sub-block is stored. This is a recording medium that stores a conversion table in which a correspondence relationship between a quantization parameter corresponding to a macroblock and a quantization parameter corresponding to the orthogonal transform block is stored for each of various orthogonal transform blocks. The conversion table storage unit 53 constitutes a conversion table storage unit.

The inverse quantization parameter generation unit 54 selects a conversion table corresponding to the type of sub-block output from the inter prediction image generation unit 58 and the block index information among the plurality of conversion tables stored in the conversion table storage unit 53. Referring to FIG. 4, a process for converting the quantization parameter corresponding to the macroblock output from the entropy decoding unit 52 into a quantization parameter corresponding to the sub-block or the orthogonal transform block is performed.
However, the dequantization parameter generation unit 54 outputs a set of table IDs and table offset values indicating the conversion table from the entropy decoding unit 52, and the types of sub-blocks output from the inter-predicted image generation unit 58 and their blocks When the index information corresponds to the table ID indicating the conversion table, the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the subblock are determined according to the change amount indicated by the table offset value. And the quantization parameter corresponding to the macroblock is converted into the quantization parameter corresponding to the sub-block according to the changed correspondence. The inverse quantization parameter generation unit 54 constitutes a quantization parameter conversion unit.

The inverse quantization unit 55 uses the quantization parameter corresponding to the sub-block or orthogonal transform block transformed by the inverse quantization parameter generation unit 54 to inverse the transform coefficient output from the entropy decoding unit 52 in units of orthogonal transform blocks. A process of quantizing and outputting a transform coefficient corresponding to the transform coefficient output from the orthogonal transform unit 17 in the image encoding device of FIG. 1 to the inverse orthogonal transform unit 56 is performed.
The inverse orthogonal transform unit 56 performs inverse orthogonal transform on the transform coefficient output from the inverse quantization unit 55, and outputs a difference image corresponding to the difference image output from the switch 16 in the image encoding device in FIG. 1 to the adder 60. Perform the process.
The inverse quantization unit 55 and the inverse orthogonal transform unit 56 constitute inverse quantization means.

When the intra prediction mode is output from the entropy decoding unit 52, the intra prediction image generation unit 57 uses the decoded image generated by the adder 60 in the intra prediction mode, by the intra prediction unit 11 in the image encoding device in FIG. A process of generating a predicted image corresponding to the generated predicted image is performed.
When the motion vector is output from the entropy decoding unit 52, the inter prediction image generation unit 58 performs a motion compensation prediction process on the decoded image stored in the frame memory 62 in units of sub-blocks, so that the image code of FIG. The process which produces | generates the prediction image corresponded to the prediction image produced | generated by the motion compensation prediction part 13 in a conversion apparatus is implemented.

If the determination result of the intra / inter determination unit 15 output from the entropy decoding unit 52 indicates that the prediction image generated by the intra prediction unit 11 is the optimal prediction image, the switch 59 generates an intra prediction image. The prediction image generated by the unit 57 is output to the adder 60, and the determination result of the intra / inter determination unit 15 indicates that the prediction image generated by the motion compensation prediction unit 13 is the optimal prediction image. For example, the process which outputs the estimated image produced | generated by the inter estimated image production | generation part 58 to the adder 60 is implemented.
The intra predicted image generation unit 57, the inter predicted image generation unit 58, and the switch 59 constitute a predicted image generation unit.

The adder 60 performs a process of adding the predicted image output from the switch 59 and the difference image output from the inverse orthogonal transform unit 56 to generate a decoded image. Note that the adder 60 constitutes a decoded image generating means.
The deblocking filter unit 61 performs a deblocking filter process on the decoded image generated by the adder 60 to compensate for distortion accompanying compression, and stores the decoded image after distortion compensation in the frame memory 62.
The frame memory 62 is a recording medium for storing the decoded image after distortion compensation.
The switch 63 performs a process of selecting the decoded image after distortion compensation by the deblocking filter unit 61 or the decoded image stored in the frame memory 62 in the display order and outputting the selected decoded image.

Next, the operation will be described.
First, the processing content of the image coding apparatus will be described.
The intra prediction unit 11 of the encoding unit 1 selects an optimal intra prediction mode for each macroblock with respect to the input image, and predicts the predicted image from the local decoded images in the vicinity that have already been encoded by the intra prediction mode. Is generated (step ST1 in FIG. 8). However, since a technique for generating a predicted image by selecting an optimal intra prediction mode is a known technique, a detailed description thereof is omitted (see, for example, H.264).
When the intra prediction unit 11 generates a prediction image from a peripheral local decoded image that has already been encoded, the intra prediction unit 11 generates a difference image between the input image and the prediction image, and outputs the difference image to the switch 16 (step ST2). . In addition, the optimal intra prediction mode is output to the entropy encoding unit 24.

The motion search unit 12 performs motion search by comparing the input image and the locally decoded image stored in the frame memory 23 in units of macro blocks, and calculates a motion vector. Since the motion vector calculation process is also a known technique, a detailed description thereof will be omitted (see, for example, H.264).
When the motion search unit 12 calculates a motion vector, the motion compensation prediction unit 13 uses the motion vector to be stored in the frame memory 23 in units of subblocks that are the same as the macroblock or smaller in size than the macroblock. A prediction image is generated by performing motion compensation prediction processing on the local decoded image (step ST3). Since the motion compensation prediction process is also a known technique, a detailed description thereof is omitted (for example, see H.264).

Here, FIG. 2 is an explanatory diagram illustrating a partition of a macroblock used when performing motion compensation prediction processing defined in H.264.
(A) shows an example in which a macroblock is composed of one partition, and the partition is a 16 × 16 block.
(B) shows an example in which a macroblock is composed of two partitions, and the two partitions are 16 × 8 blocks.
(C) shows an example in which the macroblock is composed of two partitions, and the two partitions are 8 × 16 blocks.
(D) shows an example in which a macroblock is composed of four partitions, and the four partitions are 8 × 8 blocks.
The numbers in the macroblock partitions are block indexes for specifying the positions of the partitions.

FIG. 2 is an explanatory diagram showing sub-block partitions used when performing motion compensation prediction processing defined in H.264.
(A) shows an example in which a sub-block is composed of one partition and the partition is an 8 × 8 block.
(B) shows an example in which the sub-block is composed of two partitions, and the two partitions are 8 × 4 blocks.
(C) shows an example in which the sub-block is composed of two partitions, and the two partitions are 4 × 8 blocks.
(D) shows an example in which the sub-block is composed of four partitions, and the four partitions are 4 × 4 blocks.
The numbers in the sub-block partitions are block indexes for specifying the positions of the partitions.

The motion compensation prediction unit 13 can perform motion compensation prediction that remarkably reflects the pattern even if the motion compensation prediction process is performed in units of macroblocks. If the motion compensation prediction process in units of subblock partitions as shown in FIG. 3 is also performed, the pattern can be reflected more remarkably.
Therefore, in the first embodiment, the motion compensation prediction unit 13 performs motion compensation prediction in subblock units (macroblock partition units as shown in FIG. 2 or subblock partition units as shown in FIG. 3). It is assumed that a partition smaller than a macroblock or a sub-block partition is selected as the optimum motion compensation prediction.

When the motion compensated prediction unit 13 generates a predicted image as described above, the quantization control of the coding control unit 2 indicates partition information indicating the sub-block (partition) used when performing the motion compensated prediction process. To the unit 33.
When the motion compensation prediction unit 13 generates a prediction image, the difference unit 14 generates a difference image by obtaining a difference between the input image and the prediction image, and outputs the difference image to the switch 16 (step ST4).

When the intra prediction unit 11 and the motion compensated prediction unit 13 generate a prediction image, the intra / inter determination unit 15 compares the two prediction images and determines an optimal prediction image. The most widely used method for determining the optimum predicted image is to use the value obtained by accumulating the absolute value of the difference or square of the difference of the pixel at the same pixel position in the input image and the predicted image as the evaluation value. To do. In this case, the smaller evaluation value is selected. Details are omitted.
When the intra / inter determination unit 15 determines an optimum predicted image, the intra / inter determination unit 15 outputs the predicted image to the adder 21. In addition, the determination result indicating the optimum predicted image is output to the switch 16 and the entropy encoding unit 24.

If the determination result output from the intra / inter determination unit 15 indicates that the prediction image generated by the intra prediction unit 11 is the optimal prediction image (step ST5), the switch 16 determines that the intra prediction unit 11 Is selected and output to the orthogonal transform unit 17 (step ST6).
On the other hand, if the determination result indicates that the prediction image generated by the motion compensation prediction unit 13 is an optimal prediction image (step ST5), the difference image generated by the differentiator 14 is selected and orthogonally transformed. It outputs to the part 17 (step ST7).

When the orthogonal transformation unit 17 receives the difference image from the switch 16, it orthogonally transforms the difference image in units of orthogonal transformation blocks (4 × 4 blocks or 8 × 8 blocks) as shown in FIG. The conversion coefficient of the difference image is output to the quantization unit 18 (step ST8). Further, orthogonal transform block information indicating an orthogonal transform block used when orthogonally transforming the difference image is output to the quantization control unit 33 of the encoding control unit 2.

The quantization control unit 33 of the encoding control unit 2 determines the bitstream buffer amount accumulated by the transmission buffer 25, the target code amount for each macroblock, the bitstream code amount actually generated, and the like. Then, the quantization parameter QP corresponding to the macroblock is determined (step ST9). As a process for determining the quantization parameter QP corresponding to the macroblock, for example, there is a method called TM5 used in the MPEG-2 verification test (for example, the Journal of the Television Society, April 1995, Vol49, See No4).
When determining the quantization parameter QP corresponding to the macroblock, the quantization control unit 33 converts the quantization parameter QP into a quantization parameter corresponding to the sub-block or the orthogonal transform block (step ST10).
Hereinafter, the processing content of the quantization control part 33 is demonstrated concretely.

First, the conversion table storage unit 32 of the encoding control unit 2 stores the quantization corresponding to the macroblock for each macroblock partition as shown in FIG. 2 or for each subblock partition as shown in FIG. It stores a conversion table that records the correspondence between parameters and quantization parameters corresponding to the partition. In addition, for each of the various orthogonal transform blocks, a conversion table that stores a correspondence relationship between the quantization parameter corresponding to the macro block and the quantization parameter corresponding to the orthogonal transform block is stored.

Here, FIG. 4 is an explanatory diagram showing an example of a conversion table related to a partition with a block index = 0 in a macroblock composed of two 16 × 8 blocks (see FIG. 2B).
In FIG. 4, “QP” indicates a quantization parameter corresponding to a macroblock, and “QP_part16 × 8_idx0” indicates a quantization parameter corresponding to a partition having a block index = 0.
“QPOffset_part16 × 8_idx0” is a table offset value indicating the amount of change in the correspondence between the quantization parameter QP corresponding to the macroblock and the quantization parameter QP_part16 × 8_idx0 corresponding to the partition with block index = 0. The value is transmitted to the image decoding device for each picture or slice.

When the character detection unit 31 of the encoding control unit 2 inputs an image in units of one frame or a macro block, for example, a lot of mosquito noise is generated by detecting characters included in the input image. Identifies the sub-block around the top of the character. Since various techniques have been reported for character detection processing, detailed description thereof will be omitted.
For example, when the character detection unit 31 identifies a sub-block around the upper end of a character that generates a lot of mosquito noise, the character detection unit 31 quantizes character detection information of a block index indicating the type of sub-block including the character and the position of the sub-block. Output to the control unit 33.

FIG. 5 is an explanatory diagram showing an example in which characters such as a caption superimpose are multiplexed on an input image (moving image).
In the example of FIG. 5, the letter “B” is multiplexed, and the positional relationship between the 16 × 16 macroblock (the block indicated by the bold line) and the letter “B” is shown.
In FIG. 5, for example, four 8 × 8 blocks are selected as the optimum macroblock partitions from the result of motion compensation prediction, and the upper two 8 × 8 blocks are two in each case. An example in which the block is divided into sub-blocks (8 × 4 blocks) is shown.
The numbers on the left in parentheses of each block indicate the block index that identifies the partition that constitutes the macroblock, and the numbers on the right indicate the block index that identifies the partition that constitutes the sub-block. .

It is well known that when characters are multiplexed on a moving image, mosquito noise stands out around the upper end of the characters. In the example of FIG. 5, an elliptical region indicated by a dotted line is a region where mosquito noise is conspicuous.
When this elliptical area is a flat pattern, the deterioration is more noticeably perceived by mosquito noise.
On the other hand, the same mosquito noise is generated in the lower two 8 × 8 block character portions. However, since the pattern of the character itself is fine, the deterioration is not perceived by the mosquito noise near the upper end of the character. .

Therefore, the quantization parameter corresponding to the partition of the 8 × 4 block (0, 1) (1, 1), which is a sub-block including the upper periphery of the character, is smaller than the quantization parameter corresponding to the other partition. If the table offset value is set as described above, it is possible to improve the deterioration of the sub-block including the vicinity of the upper end of the character.
However, if the quantization parameter of the macroblock itself is lowered in order to lower the quantization parameter corresponding to the subblock including the upper end of the character, the code amount in the subblocks other than the subblock including the upper end of the character is reduced. To increase. For this reason, since it is necessary to suppress the increase in the code amount, the quantization parameter corresponding to the macroblocks subsequent to the macroblock is increased, so that the image quality of other regions is deteriorated.

Therefore, in the first embodiment, only the quantization parameter corresponding to the sub-block including the vicinity of the upper end of the character is lowered, and the quantization parameter corresponding to the other sub-block is not lowered.
That is, when an image as shown in FIG. 5 is input, the character detection unit 31 outputs to the quantization control unit 33 character detection information of a block index indicating the type of sub-block including the character and the position of the sub-block. To do.
However, in the example of FIG. 1, the character detection unit 31 provides the character detection information to the quantization control unit 33, but a subblock whose image quality is deteriorated compared to other subblocks is specified. As long as the processing unit provides the quantization control unit 33 with the deterioration information of the block index indicating the type of sub-block and the position of the sub-block in which significant deterioration is detected, the processing unit other than the character detection unit 31 has a table offset. Information for changing the value may be given to the quantization control unit 33.

The quantization control unit 33 identifies a conversion table to be referred to when converting the quantization parameter corresponding to the sub-block from the plurality of conversion tables stored in the conversion table storage unit 32.
That is, if the sub-block (partition) indicated by the partition information output from the motion compensation prediction unit 13 is, for example, an 8 × 4 block (0, 1), the quantization control unit 33 uses an 8 × 4 block (0 , 1), for example, if it is 8 × 8 block (2, −), the conversion table related to 8 × 8 block (2, −) is referred to.

When the quantization control unit 33 specifies the conversion table to be referred to, based on the quantization parameter QP corresponding to the macroblock, the character detection information output from the character detection unit 31, deterioration information, and the like, as shown below. Is added with a predetermined table offset value to calculate the table index QP_table_index of the conversion table.
QP_table_index = QP + table offset value In this example, as shown in FIG. 5, in the 8 × 4 sub-block (0, 1) or (1, 1), the sub-block in which the upper end of the character is most detected is detected. When the type and its block index are assumed, for 8 × 4 sub-block (0, 1) or (1, 1), for example, a table for lowering the quantization parameter such as “−5” An offset value is determined in advance. When the sub-block indicated by the partition information output from the motion compensation prediction unit 13 is an 8 × 4 sub-block (0, 1) or (1, 1), “−5” is set as the table offset value. use. If it is a sub-block other than the sub-block, a predetermined 0-table offset value is used in order not to lower the quantization parameter.
For example, if QP = 25 and table offset value = −5, then QP_table_index = 20. For example, if QP = 25 and table offset value = 0, then QP_table_index = 25.

After calculating the table index QP_table_index of the conversion table, the quantization control unit 33 acquires the quantization parameter corresponding to the sub-block (partition) indicated by the partition information from the conversion table to be referenced using the table index QP_table_index as a key. To do.
For example, when the conversion table shown in FIG. 4 is a conversion table to be referenced, if the table index QP_table_index is “20”, the quantization parameter corresponding to the sub-block is “18”.
If the table index QP_table_index is “25”, the quantization parameter corresponding to the sub-block is “22”.

When acquiring the quantization parameter corresponding to the sub-block indicated by the partition information, the quantization control unit 33 outputs the quantization parameter to the quantization unit 18.
Further, the quantization control unit 33 refers to the table ID table of FIG. 6 stored by the conversion table storage unit 32, and refers to the conversion table that is determined in advance to use a table offset value other than “0”. A set of a table ID and a table offset value indicating the target conversion table is output to the entropy encoding unit 24 for each picture or slice.
For example, if the sub block indicated by the partition information is 8 × 4 block (0, 1), the macro block partition is 8 × 8 block, the sub block partition is 8 × 4 block, and the block index is “1”. Therefore, the table ID is “7”.

The reason why the set of the table offset value and the table ID indicating the conversion table using the table offset value is transmitted for each picture or slice is that the amount of bits is greatly increased if it is transmitted for each macroblock. The set of table ID and table offset value is multiplexed and transmitted in the picture header or slice header. The table offset value of the conversion table that has not been transmitted is used as a default value determined in advance between the encoding device and the decoding device, for example, “0”. Further, instead of sending a set of table ID and table offset value, a method of transmitting all table offset values of the conversion table for all sub-blocks in a predetermined order without assigning a table ID is also conceivable.
For each picture or slice, a flag A indicating whether or not to convert from a quantization parameter corresponding to a macroblock to a quantization parameter corresponding to a sub-block is transmitted to the image decoding apparatus side.

When transforming is performed, the above-described operation is performed. When transforming is not performed, sub-block quantization is performed using a quantization parameter corresponding to a macroblock.
At this time, the flag A may be transmitted using one bit for each macroblock, and application / non-application may be switched for each macroblock. For example, application is selected for a macroblock in which a character is detected by the character detection unit 31, and is not applied to a macroblock that has not been detected. Deterioration can be improved and adverse effects on other parts can be minimized.

Here, as a form in which the quantization control unit 33 can change the quantization parameter for the sub-block in the sub-block unit in the motion compensation prediction in the quantization unit 18, the quantization parameter corresponding to the macroblock corresponds to the sub-block. Although what is converted into a quantization parameter has been described, in the case where the quantization unit 18 can change the quantization parameter in units of orthogonal transform blocks, the macro block is referred to with reference to the transform table related to the orthogonal transform block. The quantization parameter corresponding to is converted to the quantization parameter corresponding to the orthogonal transform block.
However, since it is necessary to indicate whether to convert to a quantization parameter corresponding to a sub-block or to a quantization parameter corresponding to an orthogonal transform block, a flag B indicating which conversion is to be performed is set to a picture or slice Transmit every time.
Note that the method for converting to the quantization parameter corresponding to the orthogonal transform block is the same as the method for converting to the quantization parameter corresponding to the sub-block, and thus the description thereof is omitted.

When the quantization unit 18 receives the transform coefficient of the difference image from the orthogonal transform unit 17, the quantization unit 18 orthogonally transforms the transform coefficient of the difference image using the quantization parameter corresponding to the sub-block output from the quantization control unit 33. Quantization is performed in units of blocks (step ST11). The same quantization parameter is used for all orthogonal transform blocks included in the same sub-block.

When the quantization unit 18 quantizes the transform coefficient of the difference image, the inverse quantization unit 19 inversely quantizes the transform coefficient so that the transform coefficient corresponding to the transform coefficient output from the orthogonal transform unit 17 is inverted. Output to the orthogonal transform unit 20.
When the inverse orthogonal transform unit 20 receives the transform coefficient from the inverse quantization unit 19, the inverse orthogonal transform is performed on the transform coefficient to output a difference image corresponding to the difference image output from the switch 16 to the adder 21. .

The adder 21 adds the difference image output from the inverse orthogonal transform unit 20 and the prediction image selected by the intra / inter determination unit 15 to generate a local decoded image. The generated local decoded image is output to the deblocking filter unit 22 and is also output to the intra prediction unit 11 for generating a predicted image in intra prediction.
When the adder 21 generates a local decoded image, the deblocking filter unit 22 performs deblocking filter processing on the local decoded image to compensate for distortion due to compression, and the distortion-compensated local decoded image is stored in the frame memory 23. To store.

The entropy encoding unit 24 entropy encodes the following information to generate a bit stream (step ST12).
A conversion coefficient quantized by the quantization unit 18 A table ID and table offset value set indicating a conversion table using a table offset value (a table ID indicating a conversion table and a set of table offset values are set for each picture or slice) (Entropy encoding is performed only when the picture header or slice header is transmitted.)
The determination result of the intra / inter determination unit 15 The prediction image generation information used for generating the optimal prediction image (the prediction result generated by the intra prediction unit 11 based on the determination result output from the intra / inter determination unit 15) If the image indicates that it is an optimal prediction image, it indicates that the intra prediction mode selected by the intra prediction unit 11 and that the prediction image generated by the motion compensation prediction unit 13 is the optimal prediction image. For example, the motion vector calculated by the motion search unit 12)
Other information (for example, flags A and B, quantization parameter QP corresponding to macroblock, etc.)

When the entropy encoding unit 24 generates a bit stream, the transmission buffer 25 temporarily holds the bit stream, and then transmits the bit stream to the image decoding device via an external transmission unit such as a line. Send.

As is apparent from the above, according to the first embodiment, for each of various subblocks having a size smaller than that of the macroblock, the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the subblock. Motion compensation that is performed when a predicted image is generated from among a plurality of conversion tables stored in the conversion table storage unit 32 and a conversion table storage unit 32 that stores a conversion table that records the relationship. A quantization control unit 33 that converts a quantization parameter corresponding to the macroblock into a quantization parameter corresponding to the subblock with reference to a conversion table related to the subblock used in the prediction processing is provided. Using the quantization parameter corresponding to the sub-block transformed by the quantization control unit 33, the orthogonal transformation unit 17 Since it is configured to quantize the orthogonal coefficient of the output difference image with orthogonal transform block brings in the macroblock, the effect that can be visually image deterioration to improve the image quality of significant sub-blocks.

Next, processing contents of the image decoding apparatus will be described.
The reception buffer 51 receives the bit stream transmitted from the image encoding device in FIG. 1 and outputs the bit stream to the entropy decoding unit 52.
When receiving the bit stream from the reception buffer 51, the entropy decoding unit 52 performs entropy decoding on the bit stream (step ST21 in FIG. 9).

The entropy decoding unit 52 outputs, to the inverse quantization unit 55, the transform coefficient quantized by the quantization unit 18 of the image encoding device in FIG.
A table ID and table offset value set indicating the conversion table, flags A and B, quantization parameters corresponding to the macroblock, and the like are output to the inverse quantization parameter generation unit 54.
Further, the determination result of the intra / inter determination unit 15 is output to the switch 59.
Furthermore, if the determination result of the intra / inter determination unit 15 indicates that the prediction image generated by the intra prediction unit 11 is an optimal prediction image, the intra prediction mode that is the prediction image generation information is intra predicted. If it is output to the image generation unit 57 and the determination result of the intra / inter determination unit 15 indicates that the prediction image generated by the motion compensation prediction unit 13 is an optimal prediction image, the prediction image generation information is used. The type of a certain sub-block, its block index, and motion vector are output to the inter predicted image generation unit 58.

When the inverse quantization parameter generation unit 54 receives the quantization parameter corresponding to the set of the table ID and the table offset value, the flags A and B, and the macroblock from the entropy decoding unit 52, the flag A corresponds to the quantum corresponding to the macroblock. In the case of indicating conversion from a quantization parameter to a quantization parameter corresponding to a subblock, the conversion table storage unit 53 stores the type of subblock output from the inter prediction image generation unit 58 and its block index information. Among the plurality of conversion tables (the conversion table storage unit 53 stores the same conversion table as the conversion table storage unit 32 in the image encoding device in FIG. 1), the corresponding conversion table is specified. When a set of table IDs and table offset values is received, a sub table is generated from the quantization parameter corresponding to the macroblock by the conversion table (conversion table referred to when converting the quantization parameter) indicated by the table ID. When converting to a quantization parameter corresponding to a block, conversion is performed using the table offset value received at the same time. In the case of a conversion table in which no table offset value is transmitted, the table offset value is used as a default value common to the encoding device and the decoding device, for example, “0”.
For example, if the table ID is “7”, a conversion table having a macroblock partition of 8 × 8 blocks, a subblock partition of 8 × 4 blocks, and a block index of “1” is shown. When the type of sub-block output from the generation unit 58 and its block index information indicate this table, conversion is performed using the table offset value received simultaneously.

When the inverse quantization parameter generation unit 54 specifies the conversion table to be referred to based on the type of sub-block output from the inter-predicted image generation unit 58 and its block index information, the quantization control unit in the image encoding device in FIG. Similarly to 33, the quantization table parameter QP_table_index of the conversion table is calculated by adding the quantization parameter QP corresponding to the macroblock and the table offset value.
QP_table_index = QP + table offset value However, in the case of 8 × 4 sub-block (0, 1) or (1, 1), when converting this sub-block, for example, quantization such as “−5” A table offset value for lowering the parameter is transmitted from the image coding apparatus in FIG. 1, but if it is a sub-block other than the sub-block, the table offset value is not transmitted from the image coding apparatus in FIG. The table offset value is treated as 0 value in order not to lower the optimization parameter.
For example, if QP = 25 and table offset value = −5, then QP_table_index = 20. For example, if QP = 25 and table offset value = 0, then QP_table_index = 25.

After calculating the table index QP_table_index of the conversion table, the inverse quantization parameter generation unit 54 acquires the quantization parameter corresponding to the sub block from the conversion table using the table index QP_table_index as a key, and reverses the quantization parameter. It outputs to the quantization part 55 (step ST22).
For example, when the conversion table shown in FIG. 4 is a conversion table to be referenced, if the table index QP_table_index is “20”, the quantization parameter corresponding to the sub-block is “18”.
If the table index QP_table_index is “25”, the quantization parameter corresponding to the sub-block is “22”.
When the flag A indicates that the quantization parameter corresponding to the macroblock is not converted to the quantization parameter corresponding to the sub-block, the quantization parameter corresponding to the macroblock is output to the inverse quantization unit 55.

Here, the inverse quantization parameter generation unit 54 converts the quantization parameter corresponding to the macro block into the quantization parameter corresponding to the sub block in the sub quantization unit 55 in the sub quantization unit 55. In the inverse quantization unit 55, when the quantization parameter corresponding to the macro block is converted into the quantization parameter corresponding to the orthogonal transform block in units of the orthogonal transform block (the flag B is the quantization corresponding to the orthogonal transform block). In the case of indicating conversion to a parameter), the quantization parameter corresponding to the macroblock is converted into the orthogonal transformation block by referring to the transformation table related to the orthogonal transformation block based on the orthogonal transformation block size information output from the entropy decoding unit 52. Convert to the corresponding quantization parameter.
Note that the method for converting to the quantization parameter corresponding to the orthogonal transform block is the same as the method for converting to the quantization parameter corresponding to the sub-block, and thus the description thereof is omitted.

When the inverse quantization unit 55 receives the quantization parameter corresponding to the sub-block (or macroblock) from the inverse quantization parameter generation unit 54, the inverse quantization unit 55 uses the quantization parameter to convert the transform output from the entropy decoding unit 52. By inversely quantizing the coefficients in units of orthogonal transform blocks, transform coefficients corresponding to transform coefficients output from the orthogonal transform unit 17 in the image coding apparatus in FIG. 1 are output to the inverse orthogonal transform unit 56 (step ST23). .
Alternatively, when a quantization parameter corresponding to an orthogonal transform block is received from the inverse quantization parameter generation unit 54, the transform coefficient output from the entropy decoding unit 52 is inversely quantized in units of orthogonal transform blocks using the quantization parameter. As a result, the transform coefficient corresponding to the transform coefficient output from the orthogonal transform unit 17 in the image encoding device in FIG. 1 is output to the inverse orthogonal transform unit 56.

When receiving the transform coefficient from the inverse quantization unit 55, the inverse orthogonal transform unit 56 performs inverse orthogonal transform on the transform coefficient, thereby obtaining a difference corresponding to the difference image output from the switch 16 in the image encoding device in FIG. The image is output to the adder 60 (step ST24).

When the intra prediction image generation unit 57 receives the intra prediction mode as the prediction image generation information from the entropy decoding unit 52, the intra prediction image generation unit 57 uses the decoded image generated by the adder 60 in the intra prediction mode in the image encoding device of FIG. A prediction image corresponding to the prediction image generated by the intra prediction unit 11 is generated (step ST25).
When the inter prediction image generation unit 58 receives the motion vector as the prediction image generation information from the entropy decoding unit 52, the inter prediction image generation unit 58 performs a motion compensation prediction process on the decoded image stored in the frame memory 62 in units of sub-blocks. A prediction image corresponding to the prediction image generated by the motion compensation prediction unit 13 in the image encoding device of FIG. 1 is generated (step ST26).

If the determination result of the intra / inter determination unit 15 output from the entropy decoding unit 52 indicates that the prediction image generated by the intra prediction unit 11 is the optimal prediction image, the switch 59 (step ST27). The prediction image generated by the intra prediction image generation unit 57 is output to the adder 60 (step ST28), and the determination result of the intra / inter determination unit 15 is the optimal prediction image generated by the motion compensation prediction unit 13. If it shows that it is a prediction image (step ST27), the prediction image produced | generated by the inter prediction image production | generation part 58 will be output to the adder 60 (step ST29).

The adder 60 adds the predicted image output from the switch 59 and the difference image output from the inverse orthogonal transform unit 56 to generate a decoded image, and outputs the decoded image to the deblocking filter unit 61 ( Step ST30).
When receiving the decoded image from the adder 60, the deblocking filter unit 61 performs a deblocking filter process on the decoded image, compensates for distortion due to compression, and stores the decoded image after distortion compensation in the frame memory 62. .
The switch 63 selects the decoded image generated by the adder 60 or the decoded image stored in the frame memory 62 in the display order, and outputs the selected decoded image.

As is apparent from the above, according to the first embodiment, for each of various subblocks having a size smaller than that of the macroblock, the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the subblock. Of the plurality of conversion tables stored in the conversion table storage unit 53 storing the conversion table recording the relationship and the conversion table storage unit 53, the sub-block output from the inter prediction image generation unit 58 An inverse quantization parameter generation unit 54 that converts a quantization parameter corresponding to a macroblock into a quantization parameter corresponding to a sub-block with reference to a conversion table indicated by the type and its block index information, and an inverse quantization unit 55 corresponds to the sub-block transformed by the inverse quantization parameter generation unit 54 Since the transform coefficient output from the entropy decoding unit 52 is inversely quantized using the orthogonalization block unit by using the child parameter, the image quality of the sub-block that is visually noticeably degraded in the macroblock. There is an effect that can be improved.

In this Embodiment 1, H.264 is used. The example applied to the encoding based on H.264 has been described. In H.264, the orthogonal transform block in the macroblock is fixed to one of 4 × 4 block and 8 × 8 block.
However, in future encoding schemes, other sizes of orthogonal transform blocks such as 16 × 16, 32 × 32, 64 × 64, 128 × 128, or 16 × 8, 4 × 16, 32 × 16, etc. A plurality of types of orthogonal transform blocks may be mixed in a macro block including a rectangular orthogonal transform block.
In the first embodiment, since the image encoding device and the image decoding device hold a common conversion table for each orthogonal transform block, each orthogonal transform block is not increased so much in bit amount. Therefore, quantization can be performed with a quantization parameter different from the quantization parameter corresponding to the macroblock, and the method can be applied to a future encoding method.

In addition, when the macroblock size is expanded to 32 × 32 blocks that are larger than the current 16 × 16 blocks, it is assumed that fine patterns and flat patterns are often mixed in the same macroblock. Therefore, the ability to change the quantization parameter in units of sub-blocks is more effective in improving image quality.

Embodiment 2. FIG.
In the first embodiment, the image encoding device and the image decoding device hold a common conversion table, and the quantization parameter corresponding to the sub-block (orthogonal transform block) recorded by the common conversion table is not updated. Although a fixed one is shown, the quantization parameter corresponding to the sub-block recorded by the conversion table may be updated as necessary.

For example, in the conversion table shown in FIG. 4, 51 quantization parameters (minimum 0 to maximum) are stored in the column QP_part16 × 8_idx0 as quantization parameters corresponding to the partition (16 × 8 block) with the block index “0”. 45 quantization parameters) are recorded, but these 51 quantization parameters may be updated in units of pictures or slices.

The conversion table storage unit 32 of the image encoding device and the conversion table storage unit 53 of the image decoding device separately record the contents of 51 quantization parameters recorded in the conversion table shown in FIG. 4 as default values. .
When updating the 51 quantization parameters recorded by the conversion table, the quantization control unit 33 of the image encoding device among the plurality of conversion tables stored by the conversion table storage unit 32 is the update target. The conversion table is specified, and the 51 quantization parameters recorded by the conversion table are updated as appropriate (overwrite the updated 51 quantization parameters to the conversion table). The method for updating the quantization parameter is not particularly limited. For example, when it is necessary to change the maximum value of the quantization parameter, the quantization parameter may be updated under a user instruction.

When the quantization control unit 33 updates 51 quantization parameters, parameter update information indicating the updated 51 quantization parameters in a set with the table ID (including information for specifying a conversion table to be changed). Is output to the entropy encoding unit 24.
When receiving the parameter update information from the quantization control unit 33, the entropy encoding unit 24 performs entropy encoding by multiplexing the table ID and the parameter update information as a set in the picture header or slice header. And parameter update information is transmitted to the image decoding apparatus in units of pictures or slices.

The entropy decoding unit 52 of the image decoding apparatus performs entropy decoding on the bit stream, and outputs the table ID and parameter update information in the decoded data of the bit stream to the inverse quantization parameter generation unit 54.
When the inverse quantization parameter generation unit 54 receives the parameter update information as a set with the table ID from the entropy decoding unit 52, the inverse quantization parameter generation unit 54 refers to the parameter update information, identifies the conversion table to be changed, and according to the parameter update information, The 51 quantization parameters recorded by the conversion table are updated (overwrite 51 updated quantization parameters to the conversion table).

When the parameter update information is transmitted as a set together with the table ID, the inverse quantization parameter generation unit 54 updates the 51 quantization parameters recorded by the conversion table as described above. If update information is not transmitted as a set with the table ID, the 51 quantization parameters recorded by the conversion table are returned to the default values separately recorded at the period of the I picture or I slice. To.
When the quantization control unit 33 does not update the 51 quantization parameters recorded by the conversion table and does not transmit the parameter update information as a set with the table ID, the cycle of the I picture or I slice The 51 quantization parameters recorded by the conversion table are returned to the default values recorded separately.
Accordingly, the 51 quantization parameters in the conversion tables stored in the conversion

table storage units

32 and 53 are returned to the default values unless they are newly updated at the cycle of the I picture or I slice. .

As is apparent from the above, the quantization controller 33 updates the quantization parameter corresponding to the sub-block recorded by the conversion table, and the entropy encoder 24 updates the quantization parameter updated by the quantization controller 33. While the parameter update information indicating the content of the parameter is entropy-encoded, the entropy decoding unit 52 entropy-decodes the bitstream and outputs the parameter update information, and the inverse quantization parameter generation unit 54 is output from the entropy decoding unit 52 Since the quantization parameter corresponding to the sub-block recorded by the conversion table is updated according to the parameter update information, the conversion table held in common by the image encoding device and the image decoding device can be used as necessary. Can be updated to optimize the quantization parameter. It achieves the effect that.

Embodiment 3 FIG.
In the first and second embodiments, the conversion table storage unit 32 of the image encoding device and the conversion table storage unit 53 of the image decoding device store one conversion table for each of various sub-blocks and orthogonal transform blocks. However, a plurality of conversion tables may be stored for each of various sub-blocks and orthogonal transform blocks.
However, in this case as well, the plurality of conversion tables stored in the conversion table storage unit 32 of the image encoding device and the plurality of conversion tables stored in the conversion table storage unit 53 of the image decoding device are common conversion tables. It is.

The quantization control unit 33 of the image encoding device, when a plurality of conversion tables are stored in the conversion table storage unit 32 for each of various sub-blocks and orthogonal transform blocks, determines the resolution of the input image (horizontal pixels). Number, the number of lines in the vertical direction), the size of the macroblock (or the macroblock size for each frame when the macroblock size changes for each frame), or the encoding rate to be selected. .
Similarly to the quantization control unit 33 of the image encoding device, the inverse quantization parameter generation unit 54 of the image decoding device also stores a plurality of conversion tables for each sub-block and orthogonal transform block by the conversion table storage unit 53. If stored, the conversion table to be referred to is selected according to the resolution of the decoded image, the size of the macroblock, or the encoding rate.

For example, (1) a conversion table corresponding to a resolution of 720 pixels × 480 lines or less, (2) a conversion table corresponding to a resolution from 720 pixels × 480 lines to 1920 pixels × 1080 lines, and (3) 1920 pixels × 1080 lines. When the conversion table corresponding to the above resolution is stored in the conversion

table storage units

32 and 53, if the resolution of the input image (or decoded image) corresponds to the resolution of (1), (1) The conversion table of (2) is referred to if it corresponds to the resolution of (2), and the conversion table of (3) is referred to if it corresponds to the resolution of (3). .
When selecting a conversion table to be referred to by the quantization control unit 33 and the inverse quantization parameter generation unit 54, a parameter for specifying a conversion table to be referred to from the outside may be given for each picture or slice. The quantization control unit 33 and the inverse quantization parameter generation unit 54 may determine the resolution, macroblock size, and coding rate, and select a conversion table to be referred to.

As is apparent from the above, according to the third embodiment, when a plurality of conversion tables are stored in the conversion

table storage units

32 and 53 for each of various sub-blocks and orthogonal transform blocks, the image resolution Since the conversion table to be referred to is selected according to the macroblock size or coding rate, the optimum conversion table is referenced even if the image resolution, macroblock size or coding rate changes. As a result, the quantization parameter corresponding to the sub-block or the orthogonal transform block can be obtained.

Embodiment 4 FIG.
10 is a block diagram showing an image coding apparatus according to Embodiment 4 of the present invention. In the figure, the same reference numerals as those in FIG.
The table ID table storage unit 71 is a recording medium that stores the table ID table of FIG.
The quantization control unit 72 is similar to the quantization control unit 33 in FIG. 1, the buffer amount of the bit stream accumulated by the transmission buffer 25, the target code amount for each macroblock, and the bit stream actually generated The process of determining the quantization parameter QP corresponding to the macroblock from the code amount of the.
The quantization control unit 72 also includes a transition value (a quantization parameter corresponding to a macroblock and a quantization parameter corresponding to a sub-block used when the motion compensation prediction process is performed by the motion compensation prediction unit 13). Is input to the quantization parameter QP corresponding to the macroblock, the quantization parameter corresponding to the sub-block is calculated, and the quantization parameter is input to the quantization unit 18. Perform the output process.
Note that the quantization control unit 72 constitutes a quantization parameter calculation unit.

The entropy encoding unit 73 corresponds to the transform coefficient quantized by the quantization unit 18, the set of table ID and transition value output from the quantization control unit 72 of the encoding control unit 2, flags A and B, and the macroblock. Quantization parameter QP to be performed, the determination result output from the intra / inter determination unit 15, and the prediction image generation information used for generating the optimal prediction image (the determination result output from the intra / inter determination unit 15 is If the prediction image generated by the intra prediction unit 11 indicates that it is an optimal prediction image, the intra prediction mode selected by the intra prediction unit 11 and the prediction image generated by the motion compensation prediction unit 13 are optimal. The motion vector calculated by the motion search unit 12) is entropy-encoded and the bit stream is stored. It carries out a process of generating an over arm. The entropy encoding unit 73 constitutes entropy encoding means.

FIG. 11 is a block diagram showing an image decoding apparatus according to Embodiment 4 of the present invention. In the figure, the same reference numerals as those in FIG.
The table ID table storage unit 81 is a recording medium that stores the table ID table of FIG.
The entropy decoding unit 82 entropy-decodes the bit stream output from the reception buffer 51, and outputs the transform coefficient quantized by the quantization unit 18 of the image encoding device in FIG. 10 to the inverse quantization unit 55. The quantization parameter corresponding to the table ID and transition value set, flag, and macroblock output from the quantization control unit 72 is output to the inverse quantization parameter generation unit 83, and the determination result of the intra / inter determination unit 15 is switched to the switch 59. Execute the process to output to. In addition, if the determination result of the intra / inter determination unit 15 indicates that the prediction image generated by the intra prediction unit 11 is an optimal prediction image, the intra prediction mode that is the prediction image generation information is set to intra prediction. If it is output to the image generation unit 57 and the determination result of the intra / inter determination unit 15 indicates that the prediction image generated by the motion compensation prediction unit 13 is an optimal prediction image, the prediction image generation information is used. The process which outputs a certain motion vector to the inter estimated image production | generation part 58 is implemented. The entropy decoding unit 82 constitutes entropy decoding means.

The inverse quantization parameter generation unit 83 calculates the quantization parameter corresponding to the sub-block by adding the transition value output from the entropy decoding unit 82 to the quantization parameter QP corresponding to the macroblock, and the quantization parameter Is output to the inverse quantization unit 55. The inverse quantization parameter generation unit 83 constitutes a quantization parameter calculation unit.

In the first embodiment, the image encoding device and the image decoding device are mounted with the conversion

table storage units

32 and 53 storing the conversion table, and the quantization control unit 33 and the inverse quantization parameter generation unit 54 are An example of converting a quantization parameter corresponding to a macroblock into a quantization parameter corresponding to the subblock with reference to a conversion table related to the subblock used in the motion compensation prediction process has been described. In addition, the image decoding apparatus adds the quantization parameter QP corresponding to the macroblock and the transition value without mounting the conversion

table storage units

32 and 53 storing the conversion table, so that the quantum corresponding to the sub-block is added. An optimization parameter may be calculated.

That is, the quantization control unit 72 of the image coding apparatus determines the quantization parameter QP corresponding to the macroblock in the same manner as the quantization control unit 33 in FIG.
When the quantization control unit 72 determines the quantization parameter QP corresponding to the macroblock, as shown below, the quantization control unit 72 adds the transition value given from the outside to the quantization parameter QP corresponding to the macroblock, thereby subblock The quantization parameter corresponding to is calculated, and the quantization parameter is output to the quantization unit 18.
Quantization parameter corresponding to sub-block = QP + transition value

At this time, in order to determine for which sub-block the quantization parameter corresponding to the sub-block obtained by adding the QP value and the transition value is used, it is stored in the table ID table storage unit 71 of FIG. The table ID and transition value of the table ID table are set and output to the entropy encoding unit 73. When the transition value is not used, the transition value is set to “0”, and the QP value itself is set as a quantization parameter corresponding to the sub-block. At this time, the set of table ID and transition value is not transmitted. The set of table IDs and transition values is transmitted for each picture or slice for all sub-blocks using transition values other than “0”. However, the transition value takes a positive or negative value, and when the transition value exceeds a specified value, it is limited by the specified value.
Here, the transition value is assumed to be given from the outside, but it may be set in advance, or the quantization control unit 72 may calculate the transition value according to, for example, the degradation state of the image.
The entropy encoding unit 73 entropy-encodes the set of the table ID and the transition value output from the quantization control unit 72 instead of the table offset value to generate a bitstream.

Similar to the entropy decoding unit 52 in FIG. 7, the entropy decoding unit 82 of the image decoding apparatus receives the bit stream from the reception buffer 51 and entropy decodes the bit stream, and among the decoded data of the bit stream, FIG. The transform coefficient quantized by the quantization unit 18 of the image encoding device is output to the inverse quantization unit 55.
Further, the table ID and transition value set output from the quantization control unit 72, the flags A and B, the quantization parameter corresponding to the macroblock, and the like are output to the inverse quantization parameter generation unit 83.
Further, the determination result of the intra / inter determination unit 15 is output to the switch 59.
Furthermore, if the determination result of the intra / inter determination unit 15 indicates that the prediction image generated by the intra prediction unit 11 is an optimal prediction image, the intra prediction mode that is the prediction image generation information is intra predicted. If it is output to the image generation unit 57 and the determination result of the intra / inter determination unit 15 indicates that the prediction image generated by the motion compensation prediction unit 13 is an optimal prediction image, the prediction image generation information is used. A certain motion vector is output to the inter prediction image generation unit 58.

When the inverse quantization parameter generation unit 83 receives the quantization parameter corresponding to the set of the table ID and the transition value, the flags A and B, and the macroblock from the entropy decoding unit 82, the flag A is quantized corresponding to the macroblock. When indicating conversion from a parameter to a quantization parameter corresponding to a sub-block, the quantization value corresponding to the sub-block is added by adding the transition value to the quantization parameter QP corresponding to the macro-block as shown below. The parameter is calculated, and the quantization parameter is output to the inverse quantization unit 55.
Quantization parameter corresponding to sub-block = QP + transition value At this time, the table ID table indicates which sub-block the quantization parameter corresponding to the sub-block obtained by adding the QP value and the received transition value is used. Based on the table ID table of FIG. 6 stored in the storage unit 81, it can be specified from the received transition value and the set table ID. For a sub-block for which a set of table ID and transition value is not received, the transition value is “0” and the QP value itself is a quantization parameter corresponding to the sub-block. In this way, quantization parameters for all sub-block types and their block indexes are obtained, and the sub-block types output from the inter prediction image generation unit 58 and the quantization parameters corresponding to the block index information are reversed. It outputs to the quantization part 55.

As is apparent from the above, according to the fourth embodiment, the image encoding device and the image decoding device support macroblocks without mounting the conversion

table storage units

32 and 53 storing the conversion table. Since the quantization parameter corresponding to the sub-block is calculated by adding the quantization parameter QP and the transition value, the same effects as those of the first embodiment are obtained. Also, the difference between the quantization parameter QP corresponding to the macroblock and the quantization parameter corresponding to the sub-block is always constant, and the degree of freedom is slightly lower than that in the first embodiment, but it is necessary for transmission. The effect is that the amount of information is reduced.

Embodiment 5 FIG.
In the fifth embodiment, a case will be described in which a block that performs intra coding and a block that performs inter coding coexist in subblocks in a macroblock.
FIG. 12 is a block diagram showing an image coding apparatus according to Embodiment 5 of the present invention. In the figure, the same reference numerals as those in FIG.
Similar to the intra / inter determination unit 15 in FIG. 1, the intra / inter determination unit 91 compares the predicted image generated by the intra prediction unit 11 and the prediction image generated by the motion compensation prediction unit 13 to obtain an optimal predicted image. And the process of outputting the optimum predicted image to the adder 21 is performed.
In the previous embodiments, it was determined whether to perform intra coding or inter coding in units of macroblocks. However, in the fifth embodiment, the determination is performed for each subblock in the macroblock. . That is, the intra / inter determination unit 91 determines whether the optimal prediction image is an intra block or an inter block, based on the prediction image in units of sub-blocks generated by the intra prediction unit 11 and the motion compensation prediction unit 13. The determination result is output to the switch 16, the quantization control unit 92, and the entropy encoding unit 93.
The intra / inter determination unit 91 constitutes a block determination unit.

The quantization control unit 92 is similar to the quantization control unit 33 in FIG. 1, the buffer amount of the bit stream stored in the transmission buffer 25, the target code amount for each macroblock, and the bit stream actually generated The quantization parameter QP corresponding to the macroblock is determined from the code amount of the image, and the quantization parameter QP corresponding to the macroblock is output to the entropy encoding unit 93 at the start timing of the macroblock, and the start of the picture or slice At timing, a process of outputting a transition value QP_offset for quantization parameter conversion described below to the entropy encoding unit 93 is performed.
Also, if the quantization control unit 92 determines that the intra / inter determination unit 91 encodes the sub-block in the macroblock with the intra block, the quantization parameter conversion transition value QP_offset (transition value QP_offset is For example, it is a fixed value determined in advance according to the encoding rate or the like, or a variation value that is adaptively changed according to the buffer amount of the transmission buffer 25. The quantization parameter QP corresponding to the macroblock is converted into the quantization parameter QP_intra corresponding to the subblock, and the converted quantization parameter QP_intra is output to the quantization unit 18. Perform the process. On the other hand, when the intra / inter determination unit 91 determines that the sub-block in the macroblock is to be encoded by the inter-block, the quantization parameter QP corresponding to the macroblock is changed to the quantization parameter QP_inter that corresponds to the sub-block. To output to the quantization unit 18.
The quantization control unit 92 constitutes a quantization parameter conversion unit.

In the fifth embodiment, when a sub-block in a macro block is encoded with an inter block, the quantization control unit 92 outputs a quantization parameter QP corresponding to the macro block as a quantization parameter QP_inter corresponding to the sub-block. In this example, the quantization parameter corresponding to the sub-block determined by any of the methods described in the first to fourth embodiments may be output to the quantization unit 18.

For example, when the intra / inter determination unit 91 determines that the sub-block in the macroblock is to be encoded by the inter block, the quantization control unit 92 has a plurality of conversion tables stored by the conversion table storage unit 32. Among these, the quantization parameter QP corresponding to the macroblock is determined by referring to the conversion table related to the sub-block used in the motion compensation prediction process performed when the motion compensation prediction unit 13 generates the prediction image. The quantization parameter corresponding to the sub-block is converted, and the converted quantization parameter is output to the quantization unit 18.

For example, when the intra / inter determination unit 91 determines that the sub-block in the macroblock is to be encoded by the inter block, the quantization control unit 92 and the quantization parameter QP corresponding to the macroblock and the motion compensation prediction unit 13 is used to input a transition value with a quantization parameter corresponding to a sub-block used in the motion compensated prediction process performed when a prediction image is generated by 13, and the transition value is a quantization parameter QP corresponding to a macroblock. The quantization parameter corresponding to the sub-block is calculated, and the quantization parameter corresponding to the sub-block is output to the quantization unit 18.

The entropy encoding unit 93 includes the transform coefficient quantized by the quantization unit 18, the transition value QP_offset, the determination result of the intra / inter determination unit 91, and the prediction image generation information used to generate the optimal prediction image. (If the determination result output from the intra / inter determination unit 91 indicates that the sub-block in the macroblock is to be encoded with the intra block, the intra prediction mode and macroblock selected by the intra prediction unit 11) If it is indicated that the sub-block is to be encoded with the inter block, the motion vector calculated by the motion search unit 12) and the quantization parameter QP corresponding to the macro block (or corresponding to the macro block). Difference between the quantization parameter QP and the quantization parameter corresponding to the left macroblock ) And then entropy encodes carries out a process of generating a bitstream. The entropy encoding unit 93 constitutes entropy encoding means.

FIG. 14 is a block diagram showing an image decoding apparatus according to Embodiment 5 of the present invention. In the figure, the same reference numerals as those in FIG.
The entropy decoding unit 101 performs entropy decoding on the bit stream output from the reception buffer 51, and dequantizes the transform coefficient quantized by the quantization unit 18 of the image encoding device in FIG. The quantization parameter QP corresponding to the transition value QP_offset and the macroblock output from the quantization control unit 92 and the quantization parameter QP corresponding to the macroblock and the macroblock adjacent to the left are output to the quantization unit 55 A process of outputting the determination result of the intra / inter determination unit 91 to the switch 59, and outputting the determination result of the intra / inter determination unit 91 to the switch 59. carry out.
In addition, when the determination result of the intra / inter determination unit 91 indicates that the sub-block in the macroblock is an intra block, the entropy decoding unit 101 sets the intra prediction mode that is the prediction image generation information to the intra prediction image. When output to the generation unit 57 and the determination result of the intra / inter determination unit 91 indicates that the sub-block in the macroblock is an inter block, the motion vector that is the prediction image generation information is generated as the inter prediction image. Processing to be output to the unit 58 is performed.
Note that the entropy decoding unit 101 constitutes entropy decoding means.

When the determination result output from the entropy decoding unit 101 indicates that the determination result output from the entropy decoding unit 101 is an intra block, the inverse quantization parameter generation unit 102 performs quantization corresponding to the macroblock according to the transition value QP_offset output from the entropy decoding unit 101. The parameter QP is converted into a quantization parameter QP_intra corresponding to the subblock, the converted quantization parameter QP_intra is output to the inverse quantization unit 55, and the determination result output from the entropy decoding unit 101 is an inter block. When indicating that, the quantization parameter QP corresponding to the macroblock is output to the inverse quantization unit 55 as the quantization parameter QP_inter corresponding to the subblock. Note that the inverse quantization parameter generation unit 102 constitutes a quantization parameter conversion unit.

In the fifth embodiment, when the determination result output from the entropy decoding unit 101 indicates that the block is an inter block, the inverse quantization parameter generation unit 102 sets the quantization parameter QP corresponding to the macro block to the sub block. Although an example of outputting as the quantization parameter QP_inter will be described, the quantization parameter corresponding to the sub-block determined by any method described in the first to fourth embodiments is output to the inverse quantization unit 55 You may do it.

For example, when the determination result output from the entropy decoding unit 101 indicates that the determination result is an inter block, the inverse quantization parameter generation unit 102 performs motion compensation among a plurality of conversion tables stored in the conversion table storage unit 53. The quantization parameter QP corresponding to the macroblock is associated with the subblock with reference to the conversion table related to the subblock used in the motion compensated prediction process performed when the prediction image is generated by the prediction unit 13. The data is converted into a quantization parameter, and the converted quantization parameter is output to the inverse quantization unit 55.

For example, when the determination result output from the entropy decoding unit 101 indicates that the determination result output from the entropy decoding unit 101 is an inter block, the dequantization parameter generation unit 102 receives the quantization parameter QP and the prediction image corresponding to the macroblock from the entropy decoding unit 101. A transition value with a quantization parameter corresponding to a sub-block used in the motion compensated prediction process performed at the time of generation is input, and the transition value is added to a quantization parameter QP corresponding to a macroblock to obtain a sub-block. The quantization parameter corresponding to the block is calculated, and the quantization parameter corresponding to the sub-block is output to the inverse quantization unit 55.

Next, the operation will be described.
In the fifth embodiment, in order to describe an example in which the conversion table is not used for an intra block, the image encoding device in FIG. 12 uses the character detection unit 31, the conversion table storage unit 32, and the partition in FIG. No information is required.
Further, in the image decoding apparatus in FIG. 14, the conversion table storage unit 53 in FIG. 7 and the types of sub-blocks and signals of the block indexes (signals notified from the inter prediction image generation unit 58 to the inverse quantization parameter generation unit 54). Is unnecessary.

FIG. 13 is an explanatory diagram showing an example of an intra block in a macro block.
In FIG. 13, among the sub-blocks within the macro-block (the size is arbitrary) indicated by a bold line, the sub-blocks with hatching are blocks (intra blocks) that perform intra coding.
On the other hand, a white sub-block that is not shaded is a block (inter-block) that performs inter-coding.

The intra / inter determination unit 91 is generated by the motion compensation prediction unit 13 and an intra block that is a prediction image in units of sub-blocks generated by the intra prediction unit 11, similarly to the intra / inter determination unit 15 of FIG. 1. The inter-block which is a predicted image in sub-block units is compared to determine an optimal predicted image, and the optimal predicted image is output to the adder 21.
In addition, the intra / inter determination unit 91 outputs a determination result indicating whether the optimum predicted image is an intra block or an inter block to the switch 16, the quantization control unit 92, and the entropy encoding unit 93.

The quantization control unit 92, like the quantization control unit 33 in FIG. 1, is the buffer amount of the bit stream accumulated by the transmission buffer 25, the target code amount for each macroblock, and the bits that are actually generated. The quantization parameter QP corresponding to the macro block is determined from the code amount of the stream.
In addition, when the quantization control unit 92 receives the determination result from the intra / inter determination unit 91, when the determination result indicates that the sub-block in the macroblock is an intra block, the following equation (1) ), The quantization parameter QP corresponding to the macroblock is converted into the quantization parameter QP_intra corresponding to the sub-block in accordance with the transition value QP_offset for the quantization parameter conversion, and the quantized post-conversion The parameter QP_intra is output to the quantization unit 18.
QP_intra = QP + QP_offset (1)
However, when the converted quantization parameter QP_intra exceeds the specified value determined by the standard, the specified value is output to the quantization unit 18 as the converted quantization parameter QP_intra.

When the determination result of the intra / inter determination unit 91 indicates that the sub block in the macro block is an inter block, the quantization control unit 92 sets the quantization parameter QP corresponding to the macro block to the sub block. Is output to the quantization unit 18 as a quantization parameter QP_inter that corresponds to.
Thus, if the sub-block in the macro block is an intra block, the quantization parameter in which the transition value QP_offset for quantization parameter conversion is added to the quantization parameter QP corresponding to the macro block is If the subblock in the macroblock is an inter block, the quantization parameter QP corresponding to the macroblock is used as the quantization parameter corresponding to the subblock. It is output to the quantization unit 18.

When the quantization unit 18 receives a quantization parameter (quantization parameter QP_intra or quantization parameter QP_inter) corresponding to the sub-block from the quantization control unit 92, the quantization unit 18 uses the quantization parameter corresponding to the sub-block to perform orthogonality. The transform coefficient of the difference image output from the transform unit 17 is quantized in units of orthogonal transform blocks, and the quantized transform coefficient is output to the entropy coding unit 93.
Note that, when the quantization unit 18 quantizes the transform coefficient of the difference image using the quantization parameter QP_intra, the size of the block to be subjected to the intra coding, the prediction type of the intra coding, the transform in the intra coded block Regardless of the type of block, for the block to be intra-encoded within the same macroblock, quantization is performed with the common quantization parameter QP_intra obtained by the above equation.

The entropy encoding unit 93 is for generating a prediction image used for generating the optimal prediction image, the transform coefficient quantized by the quantization unit 18, the transition value QP_offset, the determination result of the intra / inter determination unit 91, and the like. Information (If the determination result output from the intra / inter determination unit 91 indicates that the sub-block in the macro block is an intra block, the intra prediction mode selected by the intra prediction unit 11, the macro block If the sub-block indicates that it is an inter-block, a bit stream is generated by entropy encoding the motion vector calculated by the motion search unit 12) and the quantization parameter QP corresponding to the macroblock, and the bit The stream is stored in the transmission buffer 25.
Note that the transition value QP_offset for quantization parameter conversion may be multiplexed, for example, at a predetermined position in the header information and transmitted to the image decoding apparatus in units of pictures or slices.
Alternatively, as in the first embodiment, an intra block table ID shown in FIG. 6 may be added and transmitted as a set with the table ID.

When the entropy decoding unit 101 of the image decoding apparatus stores the bit stream transmitted from the image encoding apparatus in the reception buffer 51, the entropy decoding unit 101 performs entropy decoding on the bit stream stored in the reception buffer 51.
The entropy decoding unit 101 outputs the transform coefficient quantized by the quantization unit 18 of the image encoding device in FIG. 12 among the decoded data of the bitstream to the inverse quantization unit 55 and outputs from the quantization control unit 92. The transition value QP_offset and the quantization parameter QP corresponding to the macro block and the determination result of the intra / inter determination unit 91 are output to the inverse quantization parameter generation unit 102. In addition, the determination result of the intra / inter determination unit 91 is output to the switch 59.

In addition, if the determination result of the intra / inter determination unit 91 indicates that the sub-block in the macroblock is an intra block, the entropy decoding unit 101 performs intra prediction on the intra prediction mode that is information for generating a predicted image. If output to the image generation unit 57 and the determination result of the intra / inter determination unit 91 indicates that the sub-block in the macroblock is an inter block, the motion vector that is the prediction image generation information is used as the inter prediction image. The data is output to the generation unit 58.
Note that the transition value QP_offset for quantization parameter conversion is received from the image coding apparatus in units of pictures or slices, and is thus output to the inverse quantization parameter generation unit 102 in units of pictures or slices.

When the determination result output from the entropy decoding unit 101 indicates that the determination result is an intra block, the inverse quantization parameter generation unit 102 calculates the entropy by calculating the above equation (1) as in the case of the image encoding device. In accordance with the transition value QP_offset output from the decoding unit 101, the quantization parameter QP corresponding to the macroblock is converted into the quantization parameter QP_intra corresponding to the subblock, and the converted quantization parameter QP_intra is inversely quantized. To 55.
On the other hand, when the determination result output from the entropy decoding unit 101 indicates that the block is an inter block, the quantization parameter QP corresponding to the macroblock is output to the inverse quantization unit 55 as the quantization parameter QP_inter corresponding to the subblock. To do.

When the inverse quantization unit 55 receives the quantization parameter (quantization parameter QP_intra or quantization parameter QP_inter) corresponding to the sub block from the inverse quantization parameter generation unit 102, the inverse quantization unit 55 uses the quantization parameter corresponding to the sub block. Then, the transform coefficient output from the entropy decoding unit 101 is inversely quantized in units of orthogonal transform blocks, so that the transform coefficient corresponding to the transform coefficient output from the orthogonal transform unit 17 in the image encoding device in FIG. It outputs to the orthogonal transformation part 56.

As a result, in the fifth embodiment, even if there are a mixture of intra-encoding blocks and inter-encoding blocks in the same macroblock, different quantization parameters are used with a small amount of code. Therefore, the optimum image quality can be obtained.
In the fifth embodiment, the quantization parameter QP_intra corresponding to the intra block is calculated by adding the quantization parameter QP corresponding to the macro block and the transition value QP_offset. In the first embodiment, As shown, the quantization parameter QP_intra may be obtained via a conversion table. In that case, both the image encoding device and the image decoding device need to hold the conversion table.
Needless to say, when the quantization parameter QP_intra is obtained via the conversion table, the methods described in the second and third embodiments can be applied.
Here, for the inter block, the quantization is performed using the quantization parameter QP corresponding to the macro block. However, as described above, the methods described in the first to fourth embodiments are applied. It is also possible.

In the fifth embodiment, the case where a block that performs intra coding and a block that performs inter coding are mixed in the sub-blocks in the macroblock has been described. For example, the third macroblock from the left in FIG. Thus, the present invention can also be applied to the case where the entire macroblock is encoded intra.
Hereinafter, the effect when the entire macroblock is encoded intra will be described.

When the entire macroblock is encoded intra, the quantization parameter of the intrablock (in this case, the macroblock) can be changed by changing the quantization parameter QP corresponding to the macroblock.
However, H. In a standard such as H.264, when the quantization parameter QP of the macroblock to be encoded is different from the quantization parameter of the left macroblock that has already been encoded, the difference between the quantization parameters. Need to be transmitted.
When the difference of the quantization parameter is not transmitted, the macroblock to be encoded is encoded using the quantization parameter of the macroblock on the left side.

Even when the difference is not transmitted in this way, by applying the transition value QP_offset in the fifth embodiment, the quantization parameter QP of the macroblock to be encoded (the same quantization parameter as the macroblock on the left) The quantization parameter of the intra block (in this case, the macro block) can be changed.
In this case, since the difference of the quantization parameter is not transmitted, it goes without saying that the code amount is reduced. In addition, when intra macroblocks are scattered in a frame, the effect of reducing the code amount is further increased.
The method described here can also be applied to an example of changing the intra quantization parameter after the fifth embodiment.

Embodiment 6 FIG.
FIG. 15 is a block diagram showing an image encoding apparatus according to Embodiment 6 of the present invention. In the figure, the same reference numerals as those in FIG.
The table ID table storage unit 94 stores a table ID table indicating a table ID corresponding to an intra prediction type and a divided block (a sub-block that is a unit for performing prediction within a macroblock defined for each intra prediction type). Recording medium.

When the intra / inter determination unit 91 determines that the sub-block in the macroblock is to be encoded with the intra block, the quantization control unit 95 performs the change for the quantization parameter conversion as in the quantization control unit 92 of FIG. In accordance with the value QP_offset, the quantization parameter QP corresponding to the macroblock is converted to the quantization parameter QP_intra corresponding to the subblock, and the converted quantization parameter QP_intra is output to the quantization unit 18 To do. However, unlike the quantization control unit 92 of FIG. 12, the quantization control unit 95 converts the quantization parameter corresponding to the sub-block into the quantization parameter corresponding to the intra prediction type and the divided block.
On the other hand, when the intra / inter determination unit 91 determines that the sub-block in the macroblock is to be encoded with the interblock, the quantization parameter QP corresponding to the macroblock is set in the same manner as the quantization control unit 92 in FIG. Processing to output to the quantization unit 18 as the quantization parameter QP_inter corresponding to the sub-block is performed.
The quantization control unit 95 constitutes a quantization parameter conversion unit.

The entropy encoding unit 96 includes a transform coefficient quantized by the quantization unit 18, a set of a table ID and a transition value QP_offset indicating the transform table output from the quantization control unit 95, and a quantization parameter corresponding to the macroblock. QP (or the difference value between the quantization parameter QP corresponding to the macroblock and the quantization parameter corresponding to the left macroblock), the determination result output from the intra / inter determination unit 91, and the optimal prediction Predictive image generation information used for image generation (if the determination result output from the intra / inter determination unit 91 indicates that the sub-block in the macroblock is an intra block, the intra prediction unit 11 Output intra prediction mode and intra prediction type, sub-block in macro block If shows the effect click is an inter block, it carries out a process of generating a bit stream and a motion vector) calculated by the motion search unit 12 and entropy coding. The entropy encoding unit 96 constitutes entropy encoding means.

FIG. 20 is a block diagram showing an image decoding apparatus according to Embodiment 6 of the present invention. In the figure, the same reference numerals as those in FIG.
The entropy decoding unit 103 entropy-decodes the bit stream output from the reception buffer 51, and dequantizes the transform coefficient quantized by the quantization unit 18 of the image encoding device in FIG. To the conversion unit 55. In addition, the table ID indicating the conversion table output from the quantization control unit 95 and the transition value QP_offset, and the quantization parameter QP corresponding to the macroblock (or the quantization parameter QP corresponding to the macroblock are adjacent to the left). Difference value with the quantization parameter corresponding to the macroblock of the macroblock), the determination result output from the intra / inter determination unit 91, and the intra prediction type output from the intra prediction unit 11. In addition, a process of outputting the determination result of the intra / inter determination unit 91 to the switch 59 is performed.
In addition, if the determination result of the intra / inter determination unit 91 indicates that the sub-block in the macroblock is an intra block, the entropy decoding unit 103 indicates an intra prediction type and intra prediction mode that are prediction image generation information. Is output to the intra-predicted image generation unit 57, and if the determination result of the intra / inter determination unit 91 indicates that the sub-block in the macroblock is an inter-block, a motion vector that is information for generating a predicted image is displayed. The process which outputs to the inter estimated image generation part 58 is implemented.
The entropy decoding unit 103 constitutes entropy decoding means.

When the dequantization parameter generation unit 104 indicates that the determination result output from the entropy decoding unit 103 is an intra block, the dequantization parameter generation unit 104 outputs from the entropy decoding unit 103 as in the case of the dequantization parameter generation unit 102 of FIG. A process of converting the quantization parameter QP corresponding to the macroblock into the quantization parameter QP_intra corresponding to the sub-block according to the transition value QP_offset and outputting the converted quantization parameter QP_intra to the inverse quantization unit 55 carry out. However, unlike the inverse quantization parameter generation unit 102 in FIG. 14, the inverse quantization parameter generation unit 104 converts the quantization parameter corresponding to the sub block into the quantization parameter corresponding to the intra prediction type and the divided block.
On the other hand, when the determination result output from the entropy decoding unit 103 indicates that the block is an inter block, the quantization parameter QP corresponding to the macroblock is set in the sub-block as in the inverse quantization parameter generation unit 102 of FIG. A process of outputting the corresponding quantization parameter QP_inter to the inverse quantization unit 55 is performed. Note that the inverse quantization parameter generation unit 104 constitutes a quantization parameter conversion unit.

Next, the operation will be described.
In the sixth embodiment, H.264 is used. The intra prediction type defined in H.264 will be described as an example.
FIG. 2 is an explanatory diagram illustrating an intra prediction type defined in H.264.
There are three types of intra prediction types: intra 4 × 4 prediction, intra 8 × 8 prediction, and intra 16 × 16 prediction.
In each intra prediction type, 16 × 16 macroblocks are subjected to predictive coding in units of blocks each divided into 16 4 × 4 blocks, 4 8 × 8 blocks, and 16 × 16 blocks. Is.
Each divided block is assigned a block number to indicate its position. However, the prediction encoding method itself is described in H.264. Since the contents are defined in the H.264 standard, detailed description thereof is omitted.

FIG. 17 is an explanatory diagram showing an intra prediction type and a table ID corresponding to a divided block.
When quantization is performed by changing the quantization parameter corresponding to the macroblock, the corresponding table ID and transition value QP_offset are set and transmitted to the image decoding apparatus in units of pictures or slices.
For example, when the table ID is “0”, the quantization parameter changed by the transition value QP_offset is used to quantize all the divided blocks in the intra block that is encoded by intra 4 × 4 prediction. Do. When the table ID is “3”, quantization is performed using the quantization parameter changed by the transition value QP_offset on the divided block of block number 0 in the intra block to be encoded by intra 4 × 4 prediction. Do.

FIG. 18 is an explanatory diagram illustrating an example in which quantization is performed using a quantization parameter to which the transition value QP_offset is applied to a divided block in a block to be encoded intra.
In the example of FIG. 18, the thick line frame in the figure is a 16 × 16 macroblock, and the macroblock on which a number is written is encoded intra.
In addition, the transition value QP_offset is applied to the divided blocks that are shaded.
That is, the transition value QP_offset is applied to the divided blocks of

block numbers

3 and 6 in intra 4 × 4 prediction and the divided block of block number 2 in intra 8 × 8 prediction. For divided blocks that are not shaded, the quantization parameter QP corresponding to the macroblock is used.

In this case, as shown in FIG. 17, the table IDs “6”, “9”, and “21” corresponding to the intra prediction type and the

block numbers

3, 6, and 2 are set in units of pictures or slices. The value QP_offset is transmitted to the image decoding device.
The method for converting the quantization parameter QP corresponding to the macroblock into the quantization parameter corresponding to the sub-block using the transition value QP_offset is the same as the method shown in the fifth embodiment.
Note that in the image decoding apparatus, the quantization parameter corresponding to the sub-block is applied to the divided block in which the transition value QP_offset is transmitted as a set with the table ID from the image encoding apparatus, by applying the transition value QP_offset. The quantization parameter QP corresponding to the macroblock is used for the divided block for which the transition value QP_offset has not been transmitted.

FIG. 19 is an explanatory diagram illustrating an example in which quantization is performed using a quantization parameter to which the transition value QP_offset is applied to a block to be encoded with intra.
In the example of FIG. 19, the thick line frame in FIG. It is a 32 × 32 macroblock different from H.264, and a block in which a number in the macroblock is described is encoded intra.
In addition, the intra block that is shaded applies the transition value QP_offset, and in this example, the transition value is applied to the intra 4 × 4 prediction block. For intra blocks that are not shaded, the quantization parameter QP corresponding to the macro block is used.

In this case, the transition value QP_offset is transmitted to the image decoding apparatus as a set with the table ID “0” corresponding to the intra prediction type and the block number in units of pictures or slices.
The method for converting the quantization parameter QP corresponding to the macroblock into the quantization parameter corresponding to the sub-block using the transition value QP_offset is the same as the method shown in the fifth embodiment.
Note that the image decoding apparatus applies the transition value QP_offset to the intra block in which the transition value QP_offset is transmitted as a set with the table ID from the image encoding apparatus, and sets the quantization parameter corresponding to the sub-block. The quantization parameter QP corresponding to the macroblock is used for the intra block for which the transition value QP_offset is not transmitted.
In this example, it is also possible to perform quantization using the quantization parameter to which the transition value QP_offset is applied even in the divided blocks in the intra block.

When the bit stream transmitted from the image encoding device is stored, the entropy decoding unit 103 of the image decoding device performs entropy decoding on the bit stream stored in the reception buffer 51.
The entropy decoding unit 103 outputs, to the inverse quantization unit 55, the transform coefficient quantized by the quantization unit 18 of the image encoding device in FIG. Further, a table ID indicating a conversion table output from the quantization control unit 95 and a set of transition values QP_offset, a quantization parameter QP corresponding to the macroblock, a determination result output from the intra / inter determination unit 91, The intra prediction type output from the intra prediction unit 11 is output to the inverse quantization parameter generation unit 104, and the determination result of the intra / inter determination unit 91 is output to the switch 59.

In the inverse quantization parameter generation unit 104, the determination result output from the entropy decoding unit 103 is an intra block, and the intra prediction type of the intra block and the divided block number on which the inverse quantization process is performed indicate an intra prediction type. And the block number, the quantization parameter corresponding to the macroblock is changed to the quantization parameter corresponding to the subblock (intra prediction type and block number) according to the transition value QP_offset output from the entropy decoding unit 103. Corresponding quantization parameter), and outputs the converted quantization parameter to the inverse quantization unit 55.
Note that the inverse quantization parameter generation unit 104 uses the intra prediction type and block number for which the transition value QP_offset is not transmitted from the image encoding device even when the determination result output from the entropy decoding unit 103 indicates that it is an intra block. In some cases, the quantization parameter QP corresponding to the macroblock is output to the inverse quantization unit 55 as the quantization parameter corresponding to the subblock.
On the other hand, when the determination result output from the entropy decoding unit 103 indicates that the block is an inter block, the quantization parameter QP corresponding to the macroblock is output to the inverse quantization unit 55 as the quantization parameter corresponding to the subblock. .

As is apparent from the above, according to the sixth embodiment, encoding is performed using different quantization parameters with a small amount of code for each small block of intra blocks or intra prediction types. Since it can be performed, an optimal image quality can be obtained.
In the sixth embodiment, the quantization parameter corresponding to the intra block is obtained by adding the quantization parameter corresponding to the macro block and the transition value. However, as in the first embodiment, the quantization parameter is obtained via the conversion table. Thus, the quantization parameter corresponding to the intra block may be obtained.

In that case, both the image encoding device and the image decoding device need to hold the conversion table. Needless to say, when the quantization parameter corresponding to the intra block is obtained via the conversion table, the methods shown in the second and third embodiments can be applied.
Here, although the quantization is performed on the inter block with the quantization parameter corresponding to the macro block, the method described in the first to fourth embodiments can be applied.

Embodiment 7 FIG.
FIG. 21 is a block diagram showing an image coding apparatus according to Embodiment 7 of the present invention. In the figure, the same reference numerals as those in FIG.
The table ID table storage unit 97 includes an orthogonal transform mode for an intra block (orthogonal transform mode in a block that performs intra coding in a macroblock) and an orthogonal transform block (a block that performs orthogonal transform defined for each orthogonal transform mode). ) Is a recording medium storing a table ID table indicating a table ID corresponding to.

When the intra / inter determination unit 91 determines that the sub-block in the macroblock is to be encoded with the intra block, the quantization control unit 98 performs the quantization parameter conversion as in the quantization control unit 92 of FIG. The quantization parameter QP corresponding to the macroblock is converted into the quantization parameter QP_intra corresponding to the sub-block in accordance with the transition value QP_offset of, and the converted quantization parameter QP_intra is output to the quantization unit 18 To implement. However, unlike the quantization control unit 92 in FIG. 12, the quantization control unit 98 sets the quantization parameter corresponding to the sub block to the quantization parameter corresponding to the orthogonal transform mode of the intra block and the block number to be orthogonal transformed. Convert.
On the other hand, when the intra / inter determination unit 91 determines that the sub-block in the macroblock is to be encoded with the inter block, the quantization parameter QP corresponding to the macroblock is set as in the quantization control unit 92 of FIG. A process of outputting the quantization parameter QP_inter that corresponds to the sub-block to the quantization unit 18 is performed.
The quantization control unit 98 constitutes a quantization parameter conversion unit.

The entropy encoding unit 99 includes a transform coefficient quantized by the quantization unit 18, a set of a table ID and a transition value QP_offset indicating the transform table output from the quantization control unit 98, and a quantization parameter corresponding to the macroblock. QP (or the difference value between the quantization parameter QP corresponding to the macroblock and the quantization parameter corresponding to the left macroblock), the determination result output from the intra / inter determination unit 91, and the optimal prediction Predictive image generation information used for image generation (if the determination result output from the intra / inter determination unit 91 indicates that the sub-block in the macroblock is an intra block, the intra prediction unit 11 Output intra prediction mode, intra prediction type and orthogonal transform mode, macro block If shows the effect sub-block in the click is an inter block, it carries out a process of generating a bit stream and a motion vector) calculated by the motion search unit 12 and entropy coding. The entropy encoding unit 99 constitutes entropy encoding means.

FIG. 24 is a block diagram showing an image decoding apparatus according to Embodiment 7 of the present invention. In the figure, the same reference numerals as those in FIG.
The entropy decoding unit 105 entropy-decodes the bit stream output from the reception buffer 51, and dequantizes the transform coefficient quantized by the quantization unit 18 of the image encoding device in FIG. To the conversion unit 55. Also, the table ID indicating the conversion table output from the quantization control unit 98 and the transition value QP_offset, and the quantization parameter QP corresponding to the macroblock (or the quantization parameter QP corresponding to the macroblock are adjacent to the left). The difference between the quantization parameter corresponding to the macroblock), the determination result output from the intra / inter determination unit 91, and the intra prediction type and orthogonal transform mode output from the intra prediction unit 11. A process of outputting to the parameter generation unit 106 and outputting the determination result of the intra / inter determination unit 91 to the switch 59 is performed.
Also, if the determination result of the intra / inter determination unit 91 indicates that the sub-block in the macroblock is an intra block, the entropy decoding unit 105 sets the intra prediction mode that is the prediction image generation information to the intra prediction image. When output to the generation unit 57 and the determination result of the intra / inter determination unit 91 indicates that the sub-block in the macroblock is an inter block, the motion vector that is the prediction image generation information is generated as the inter prediction image. Processing to be output to the unit 58 is performed.
The entropy decoding unit 105 constitutes entropy decoding means.

When the determination result output from the entropy decoding unit 105 indicates that the determination result is an intra block, the inverse quantization parameter generation unit 106 outputs the same from the entropy decoding unit 105 as in the case of the inverse quantization parameter generation unit 102 in FIG. Processing for converting the quantization parameter QP corresponding to the macroblock into the quantization parameter QP_intra corresponding to the sub-block according to the transition value QP_offset and outputting the converted quantization parameter QP_intra to the inverse quantization unit 55 carry out. However, unlike the inverse quantization parameter generation unit 102 in FIG. 14, the inverse quantization parameter generation unit 106 has a quantization parameter corresponding to an intra block orthogonal transform mode and an orthogonal transform block number as a quantization parameter corresponding to a sub-block. Convert to
On the other hand, when the determination result output from the entropy decoding unit 105 indicates that the block is an interblock, the quantization parameter QP corresponding to the macroblock is set in the subblock as in the case of the inverse quantization parameter generation unit 102 in FIG. A process of outputting the corresponding quantization parameter QP_inter to the inverse quantization unit 55 is performed. Note that the inverse quantization parameter generation unit 106 constitutes a quantization parameter conversion unit.

Next, the operation will be described.
FIG. 22 is an explanatory diagram showing an orthogonal transform mode of an intra block.
As an orthogonal transform mode of an intra block, H.264 is used. In H.264, 4 × 4 block conversion and 8 × 8 block conversion are defined.
In the 4 × 4 block transform, a 16 × 16 macroblock is divided into 16 4 × 4 blocks, and orthogonal transform is performed for each 4 × 4 block.
In the 8 × 8 block transform, the 8 × 8 block is divided into four 8 × 8 blocks, and orthogonal transform is performed for each 8 × 8 block. H. In 16 × 16 block transformation not defined by H.264, orthogonal transformation is performed with one 16 × 16 block.
Each divided orthogonal transform block is assigned a block number to indicate its position.

H. H.264 stipulates that 4 × 4 block conversion is performed in the intra 4 × 4 prediction type and 8 × 8 block conversion is performed in the intra 8 × 8 prediction type, but in the seventh embodiment, intra prediction is performed. Regardless of type, H. It is assumed that the orthogonal transform mode can be arbitrarily selected including those not defined by H.264.

FIG. 23 is an explanatory diagram showing an orthogonal transform mode of an intra block and a table ID corresponding to the orthogonal transform block.
When quantization is performed by changing the quantization parameter corresponding to the macroblock, the corresponding table ID and transition value QP_offset are set and transmitted to the image decoding apparatus in units of pictures or slices.
For example, when the table ID is “0”, quantization is performed on all orthogonal transform blocks subjected to orthogonal transform in the 4 × 4 block transform mode, using the quantization parameter changed with the transition value QP_offset. . When the table ID is “3”, quantization is performed using the quantization parameter changed by the transition value QP_offset with respect to the orthogonal transformation block of block number 0 for which orthogonal transformation is performed in the 4 × 4 block transformation mode. .

The intra prediction type shown in FIG. 18 and FIG. 19 shown in the sixth embodiment is replaced with the orthogonal transform mode, and the divided block number is replaced with the orthogonal transform block number. Since the operation is the same as the operation of the image coding apparatus in the sixth embodiment, detailed description is omitted here.

When the bit stream transmitted from the image encoding device is stored, the entropy decoding unit 105 of the image decoding device performs entropy decoding on the bit stream stored in the reception buffer 51.
The entropy decoding unit 105 outputs, to the inverse quantization unit 55, the transform coefficient quantized by the quantization unit 18 of the image encoding device in FIG. In addition, a table ID indicating a conversion table output from the quantization control unit 98 and a set of transition values QP_offset, a quantization parameter QP corresponding to the macroblock, a determination result output from the intra / inter determination unit 91, The intra prediction type and the orthogonal transform mode output from the intra prediction unit 11 are output to the inverse quantization parameter generation unit 106, and the determination result of the intra / inter determination unit 91 is output to the switch 59.

When the determination result output from the entropy decoding unit 105 indicates that the determination result is an intra block, the inverse quantization parameter generation unit 106 determines the quantum corresponding to the macroblock according to the transition value QP_offset output from the entropy decoding unit 105. The quantization parameter is converted to a quantization parameter (quantization parameter corresponding to the orthogonal transform mode and orthogonal transform block number of the intra block) corresponding to the sub-block, and the converted quantization parameter is output to the inverse quantization unit 55. .
On the other hand, when the determination result output from the entropy decoding unit 105 indicates that the block is an inter block, the quantization parameter QP corresponding to the macroblock is output to the inverse quantization unit 55 as the quantization parameter corresponding to the subblock. .
Note that, even when the determination result output from the entropy decoding unit 105 indicates that the determination result is an intra block, the inverse quantization parameter generation unit 106 does not transmit a transition value from the image encoding device, and the quantum value corresponding to the macro block is determined. The quantization parameter is output to the inverse quantization unit 55 as a quantization parameter corresponding to the subblock.

As is apparent from the above, according to the seventh embodiment, encoding is performed using different quantization parameters with a small amount of code for each orthogonal transform mode of an intra block or each orthogonal transform block. Therefore, the optimum image quality can be obtained.
In the seventh embodiment, the quantization parameter corresponding to the orthogonal transform mode or the orthogonal transform block is obtained by adding the quantization parameter corresponding to the macroblock and the transition value. Similarly, a quantization parameter corresponding to an orthogonal transform mode or an orthogonal transform block may be obtained via a transform table.

In that case, both the image encoding device and the image decoding device need to hold the conversion table. Needless to say, the methods shown in the second and third embodiments can be applied to obtain the orthogonal transformation mode or the quantization parameter corresponding to the orthogonal transformation block via the transformation table.
In addition, the contents described in the seventh embodiment are orthogonal transform blocks having a size of 32 × 32, 64 × 64, 128 × 128 or the like, or rectangular orthogonal shapes such as 16 × 8, 4 × 16, and 32 × 16. The same applies to the conversion block.
Further, in the seventh embodiment, the quantization parameter QP corresponding to the macroblock is subjected to the orthogonal transformation mode and orthogonal transformation of the intra block according to the quantization parameter transformation transition value QP_offset specified by the table ID. In the image encoding apparatus and the image decoding apparatus, a predetermined common quantum is determined for each of the intra block orthogonal transform mode and the block number to be subjected to orthogonal transform. By retaining the transition value QP_offset (default value) for quantization parameter conversion, the quantization parameter QP corresponding to the macroblock is converted without transmitting the table ID and the transition value QP_offset for quantization parameter conversion. You can also.
Here, although the quantization is performed on the inter block with the quantization parameter corresponding to the macro block, the method described in the first to fourth embodiments can be applied.

Since the image encoding device, the image decoding device, the image encoding method, and the image decoding method according to the present invention can improve the image quality of the sub-block that is visually noticeably deteriorated in the macroblock, The present invention is suitable for use in an image encoding device and an image encoding method for encoding, an image decoding device and an image decoding method for decoding an image encoded by the image encoding device.

Claims

Differential image generation means for encoding the input image in units of macroblocks, generating a prediction image from the locally decoded image after encoding, and generating a difference image between the input image and the prediction image; For each of the various sub-blocks having a size smaller than that of the macro block, a conversion table that records the correspondence between the quantization parameter corresponding to the macro block and the quantization parameter corresponding to the sub block is stored. Of the plurality of conversion tables stored in the conversion table storage means and the conversion table storage means, sub-blocks used in motion compensated prediction processing performed when a prediction image is generated by the difference image generation means The quantization parameter corresponding to the macroblock is assigned to the subblock with reference to the conversion table according to A quantization parameter conversion unit that converts the quantization parameter to a quantization parameter to be orthogonally converted from the difference image generated by the difference image generation unit, and a quantization parameter corresponding to the sub-block converted by the quantization parameter conversion unit is used. A quantization means for quantizing the transform coefficient of the difference image, a transform coefficient quantized by the quantization means, and a prediction image generation used when the prediction image is generated by the difference image generation means. An image encoding apparatus comprising: entropy encoding means for entropy encoding information and a quantization parameter corresponding to the macroblock to generate a bitstream.
In addition to the conversion table in which the correspondence relationship between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the sub-block is recorded, the conversion table storage unit performs the orthogonal transform of the difference image by the quantization unit. For each of the various orthogonal transformation blocks used in the above, a transformation table that records the correspondence between the quantization parameter corresponding to the macro block and the quantization parameter corresponding to the orthogonal transformation block is stored.
The quantization parameter conversion means refers to a conversion table related to an orthogonal transformation block used when the difference image is orthogonally transformed by the quantization means among the plurality of transformation tables stored by the transformation table storage means. Then, the quantization parameter corresponding to the macroblock is converted into the quantization parameter corresponding to the orthogonal transform block,
The quantization means quantizes the transform coefficient of the difference image using the quantization parameter corresponding to the orthogonal transform block instead of the quantization parameter corresponding to the sub-block transformed by the quantization parameter transform means. The image encoding apparatus according to claim 1, wherein
When a table offset value indicating the amount of change in the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the sub-block is given, the quantization parameter conversion means is configured to execute the macroblock according to the table offset value. Change the correspondence between the quantization parameter corresponding to and the quantization parameter corresponding to the sub-block, and convert the quantization parameter corresponding to the macroblock to the quantization parameter corresponding to the sub-block according to the changed correspondence And
The entropy encoding means sets the table offset value and a table ID indicating a conversion table in which the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the sub-block is changed according to the table offset value The image according to claim 1, wherein the transform coefficient quantized by the quantization unit and the prediction image generation information used when the prediction image is generated by the difference image generation unit are entropy-encoded. Encoding device.
The quantization parameter conversion means updates the quantization parameter corresponding to the sub-block recorded by the conversion table,
2. The image coding apparatus according to claim 1, wherein the entropy coding means performs entropy coding on parameter update information indicating the content of the quantization parameter updated by the quantization parameter conversion means.
When a plurality of conversion tables are stored in the conversion table storage unit for each of the various sub-blocks, the quantization parameter conversion unit converts the conversion to be referred to according to the resolution of the input image, the size of the macroblock, or the encoding rate. 2. The image encoding apparatus according to claim 1, wherein a table is selected.
Differential image generation means for encoding the input image in units of macroblocks, generating a prediction image from the locally decoded image after encoding, and generating a difference image between the input image and the prediction image; A transition value between a quantization parameter corresponding to a macroblock and a quantization parameter corresponding to a sub-block used in a motion compensation prediction process performed when a prediction image is generated by the difference image generation unit. The quantization parameter calculation means for calculating the quantization parameter corresponding to the sub-block by adding the transition value to the quantization parameter corresponding to the macroblock, and the difference image generated by the difference image generation means are orthogonal Using the quantization parameter corresponding to the sub-block calculated by the quantization parameter calculation means, Quantization means for quantizing the transform coefficient of the image, the transform coefficient quantized by the quantization means, and the transition value added to the quantization parameter corresponding to the macroblock, and the quantization corresponding to the subblock A parameter is calculated, and a set of an ID and the transition value for specifying the sub-block quantized using the quantization parameter is used when a predicted image is generated by the difference image generation unit. And an entropy encoding means for entropy encoding the prediction image generation information and the quantization parameter corresponding to the macroblock to generate a bitstream.
Entropy decodes the bitstream, and outputs the quantized transform coefficient that is the decoded data of the bitstream, the prediction image generation information used when the prediction image is generated, and the quantization parameter corresponding to the macroblock Stores the entropy decoding means and a conversion table recording the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the subblock for each of the various subblocks having a size smaller than the macroblock. Among the plurality of conversion tables stored in the conversion table storage means and the conversion table storage means, the motion compensated prediction processing is performed in the prediction image generation information used when the prediction image is generated. The entropy recovery is performed with reference to the conversion table for the used sub-block. Quantization parameter conversion means for converting a quantization parameter corresponding to the macroblock output from the means into a quantization parameter corresponding to the subblock, and a quantization parameter corresponding to the subblock converted by the quantization parameter conversion means And inverse quantization means for inversely transforming the transform coefficient output from the entropy decoding means and inversely transforming the transform coefficient after the inverse quantization, and prediction image generation information output from the entropy decoding means Is used to generate a decoded image by adding a prediction image generating unit that generates a prediction image, a prediction image generated by the prediction image generation unit, and a difference image that is an inverse orthogonal transformation result of the inverse quantization unit An image decoding apparatus comprising decoded image generation means.
In addition to the conversion table in which the correspondence relationship between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the sub-block is recorded, the conversion table storage unit converts the conversion coefficient after the inverse quantization by the inverse quantization unit. For each of various orthogonal transform blocks used when the inverse orthogonal transform is performed, a conversion table that records the correspondence between the quantization parameter corresponding to the macro block and the quantization parameter corresponding to the orthogonal transform block And store
The quantization parameter transforming means is an orthogonal transform used when the transform coefficient after inverse quantization is inversely orthogonal transformed by the inverse quantization means among the plurality of transform tables stored by the transform table storing means. Referring to the conversion table for the block, the quantization parameter corresponding to the macroblock is converted to the quantization parameter corresponding to the orthogonal transform block,
The inverse quantization means is output from the entropy decoding means using the quantization parameter corresponding to the orthogonal transform block instead of the quantization parameter corresponding to the sub-block transformed by the quantization parameter transform means. 8. The image decoding apparatus according to claim 7, wherein the transformed coefficient is inversely quantized.
The entropy decoding means performs entropy decoding on the bitstream, and outputs a table offset value indicating the amount of change in the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the sub-block,
The quantization parameter conversion means changes the correspondence relationship between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the subblock according to the table offset value, and supports the macroblock according to the changed correspondence relationship. The image decoding apparatus according to claim 7, wherein the quantization parameter to be converted is converted into a quantization parameter corresponding to the sub-block.
The entropy decoding means entropy-decodes the bitstream, and uses the table offset value indicating the amount of change in the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the sub-block, and the above table offset value To output a set with a table ID for identifying a sub-block whose correspondence is to be changed,
The quantization parameter converting means changes the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the subblock according to the table offset value for the subblock specified by the table ID. The image decoding apparatus according to claim 7, wherein the quantization parameter corresponding to the macroblock is converted into a quantization parameter corresponding to the sub-block according to the changed correspondence.
The entropy decoding means entropy decodes the bitstream, and outputs parameter update information indicating the update contents of the quantization parameter corresponding to the sub-block recorded by the conversion table,
8. The image decoding apparatus according to claim 7, wherein the conversion table storage unit updates the quantization parameter corresponding to the sub-block recorded by the conversion table according to the parameter update information.
When a plurality of conversion tables are stored in the conversion table storage unit for each of various sub-blocks, the quantization parameter conversion unit converts the reference to be referred to according to the resolution of the decoded image, the size of the macroblock, or the encoding rate. 8. The image decoding apparatus according to claim 7, wherein a table is selected.
The bitstream is entropy-decoded to generate a quantized transform coefficient, which is the decoded data of the bitstream, a transition value between a quantization parameter corresponding to a macroblock and a quantization parameter corresponding to a subblock, and a predicted image. Entropy decoding means for outputting the prediction image generation information and the quantization parameter corresponding to the macroblock used in the process, and the transition value output from the entropy decoding means is added to the quantization parameter corresponding to the macroblock Output from the entropy decoding means using the quantization parameter calculating means for calculating the quantization parameter corresponding to the sub-block and the quantization parameter corresponding to the sub-block calculated by the quantization parameter calculating means. Dequantize the transform coefficient and change it after dequantization. Inverse quantization means for inverse orthogonal transform of coefficients, prediction image generation means for generating a prediction image using the prediction image generation information output from the entropy decoding means, and prediction generated by the prediction image generation means An image decoding apparatus comprising: a decoded image generation unit configured to generate a decoded image by adding an image and a difference image which is a result of inverse orthogonal transform of the inverse quantization unit.
The bitstream is entropy decoded, and the quantized transform coefficient, which is the decoded data of the bitstream, the transition value between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the sub-block, and the transition value are A set with a table ID for specifying a sub-block for which a quantization parameter corresponding to a sub-block is calculated by adding to a quantization parameter corresponding to a macro-block, used when a predicted image is generated Entropy decoding means for outputting the prediction image generation information and the quantization parameter corresponding to the macroblock, and for the sub-block specified by the table ID, the transition value output from the entropy decoding means is used as the macroblock. Is added to the quantization parameter corresponding to A transform parameter output from the entropy decoding means using a quantization parameter calculating means for calculating a quantization parameter corresponding to a sub-block and a quantization parameter corresponding to a sub-block calculated by the quantization parameter calculating means. Inverse quantization, inverse quantization means for inverse orthogonal transform of the transform coefficient after inverse quantization, prediction image generation means for generating a prediction image using the prediction image generation information output from the entropy decoding means, An image decoding apparatus comprising: a decoded image generating unit that generates a decoded image by adding a predicted image generated by the predicted image generating unit and a difference image that is a result of inverse orthogonal transformation of the inverse quantization unit.
The difference image generation means encodes the input image in units of macroblocks, generates a prediction image from the locally decoded image after encoding, and generates a difference image between the input image and the prediction image The correspondence relationship between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the subblock is recorded for each of the difference image generation processing step and the various subblocks having a size smaller than the macroblock. Referring to the conversion table related to the sub-block used in the motion compensation prediction process performed when the prediction image is generated in the difference image generation processing step, the quantization parameter conversion unit performs the macro parameter conversion process. A quantization parameter for converting a quantization parameter corresponding to a block into a quantization parameter corresponding to the sub-block. Data conversion processing step, and the quantization means orthogonally transforms the difference image generated in the difference image generation processing step, using the quantization parameter corresponding to the sub-block converted in the quantization parameter conversion processing step. A quantization processing step for quantizing the transform coefficient of the difference image, a transform coefficient quantized by the entropy coding means in the quantization processing step, and a prediction image generated in the difference image generation processing step. An image encoding method comprising: entropy encoding processing step for entropy encoding a used prediction image generation information and a quantization parameter corresponding to the macroblock to generate a bitstream.
The difference image generation means encodes the input image in units of macroblocks, generates a prediction image from the locally decoded image after encoding, and generates a difference image between the input image and the prediction image A difference image generation process step, a quantization parameter calculation unit that uses a quantization parameter corresponding to a macroblock, and a motion compensation prediction process performed when a prediction image is generated in the difference image generation process step. Quantization parameter calculation processing step of inputting a transition value with a quantization parameter corresponding to a block, and adding the transition value to a quantization parameter corresponding to the macroblock to calculate a quantization parameter corresponding to the sub-block And a quantizing unit orthogonally transforms the difference image generated in the difference image generation processing step to obtain the quantization parameter. The quantization processing step for quantizing the transform coefficient of the difference image using the quantization parameter corresponding to the sub-block calculated in the data calculation processing step, and the entropy encoding means are quantized in the quantization processing step. The quantization coefficient corresponding to the sub block is calculated by adding the transform coefficient and the transition value to the quantization parameter corresponding to the macro block, and the sub block quantized using the quantization parameter is calculated. Entropy coding of a set of IDs for identification and the transition values, prediction image generation information used when a prediction image is generated in the difference image generation processing step, and quantization parameters corresponding to the macroblocks And an entropy encoding processing step for generating a bitstream.
The entropy decoding means entropy decodes the bitstream, and the quantized transform coefficient that is the decoded data of the bitstream, the prediction image generation information used when the prediction image is generated, and the quantum corresponding to the macroblock The entropy decoding processing step for outputting the quantization parameter and the correspondence relationship between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the subblock are recorded for each of various subblocks having a size smaller than the macroblock. Quantization parameter conversion means with reference to the conversion table related to the sub-block used in the motion compensation prediction processing in the prediction image generation information used when the prediction image is generated Is output to the macroblock output in the above entropy decoding process step. A quantization parameter conversion processing step for converting the corresponding quantization parameter into a quantization parameter corresponding to the sub-block, and a quantization parameter corresponding to the sub-block converted by the inverse quantization means in the quantization parameter conversion processing step. And using the inverse quantization process step for inversely transforming the transform coefficient output in the entropy decoding process step and performing the inverse orthogonal transform on the transform coefficient after the inverse quantization, and the predictive image generation means outputting in the entropy decoding process step. A predicted image generation processing step for generating a predicted image using the predicted image generation information, and a decoded image generated by the decoded image generation means in the predicted image generation processing step and an inverse orthogonal in the inverse quantization processing step. A decoded image generation processing step of generating a decoded image by adding the difference images as the conversion results; For example was the image decoding method.
Entropy decoding means entropy-decodes the bitstream, quantized transform coefficients that are decoded data of the bitstream, transition values between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the subblock, and prediction An entropy decoding processing step for outputting the prediction image generation information used when the image is generated and a quantization parameter corresponding to the macroblock, and a transition value output by the quantization parameter calculation means in the entropy decoding processing step Is added to the quantization parameter corresponding to the macroblock to calculate a quantization parameter corresponding to the sub-block, and the sub-quantization means calculates the sub-parameter calculated in the quantization parameter calculation processing step. Quantization parameters corresponding to blocks And the inverse quantization process step of inversely transforming the transform coefficient output in the entropy decoding process step and inverse orthogonal transform of the transform coefficient after the inverse quantization, and the predicted image generation means is the entropy decoding process step A prediction image generation processing step for generating a prediction image using the prediction image generation information output in step (b), a prediction image generated by the decoded image generation means in the prediction image generation processing step, and the inverse quantization processing step. An image decoding method comprising: a decoded image generation processing step of generating a decoded image by adding difference images as inverse orthogonal transformation results.
Differential image generation means for encoding the input image in units of macroblocks, generating a prediction image from the locally decoded image after encoding, and generating a difference image between the input image and the prediction image; For each of the various sub-blocks having a size smaller than that of the macro block, a conversion table that records the correspondence between the quantization parameter corresponding to the macro block and the quantization parameter corresponding to the sub block is stored. Of the plurality of conversion tables stored in the conversion table storage means and the conversion table storage means, the sub-blocks used in the intra prediction process performed when the prediction image is generated by the difference image generation means Referring to the conversion table, the quantization parameter corresponding to the macroblock is assigned to the subblock. A quantization parameter conversion unit that converts the quantization parameter to a quantization parameter to be orthogonally converted from the difference image generated by the difference image generation unit, and a quantization parameter corresponding to the sub-block converted by the quantization parameter conversion unit is used. A quantization means for quantizing the transform coefficient of the difference image, a transform coefficient quantized by the quantization means, and a prediction image generation used when the prediction image is generated by the difference image generation means. An image encoding apparatus comprising: entropy encoding means for entropy encoding information and a quantization parameter corresponding to the macroblock to generate a bitstream.
Differential image generation means for encoding the input image in units of macroblocks, generating a prediction image from the locally decoded image after encoding, and generating a difference image between the input image and the prediction image; A block determination unit for determining whether to perform encoding as an intra block or an inter block for each of various sub-blocks having a size smaller than that of the macro block, and a sub-block within the macro block by the block determination unit. If it is determined that the block is to be encoded as an intra block, the quantization parameter corresponding to the macroblock is converted into the quantization parameter corresponding to the sub-block according to the transition value for quantization parameter conversion. The converted quantization parameter is output, and the block determination means A quantization parameter conversion unit that outputs a quantization parameter corresponding to the macroblock as a quantization parameter corresponding to the subblock when it is determined that encoding is performed using an inner subblock as an interblock; and the difference image A quantization unit that orthogonally transforms the difference image generated by the generation unit, and that quantizes the transform coefficient of the difference image using a quantization parameter corresponding to the sub-block output from the quantization parameter conversion unit; The transform coefficient quantized by the quantization means, the transition value, the determination result of the block determination means, the prediction image generation information used when the prediction image is generated by the difference image generation means, and the macroblock An entropy-encoded quantization parameter corresponding to The image coding apparatus that includes a copy encoding means.
For each of the various sub-blocks, conversion table storage means is provided for storing a conversion table that records the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the sub-block. If
When the quantization parameter converting unit determines that the sub-block in the macroblock is to be encoded as the intra block by the block determining unit, the difference image is selected from the plurality of conversion tables stored by the conversion table storing unit. The quantization parameter corresponding to the macroblock is quantized with reference to the conversion table related to the subblock used in the intra prediction processing performed when the prediction image is generated by the generation unit. 21. The image encoding apparatus according to claim 20, wherein the image encoding apparatus converts the parameter into a parameter and outputs the converted quantization parameter.
The quantization parameter conversion unit is configured to generate a prediction image by the quantization parameter corresponding to the macroblock and the difference image generation unit when the block determination unit determines to encode the sub-block in the macroblock as an intra block. The shift value with the quantization parameter corresponding to the sub-block used in the intra prediction processing performed in the step is input, and the shift value is added to the quantization parameter corresponding to the macro-block to correspond to the sub-block. The image coding apparatus according to claim 20, wherein a quantization parameter is calculated and a quantization parameter corresponding to the sub-block is output.
The quantization parameter conversion means converts the quantization parameter corresponding to the intra prediction type as the quantization parameter corresponding to the sub-block when the block determination means determines that the sub-block in the macro block is encoded as the intra block. 21. The image encoding apparatus according to claim 20, wherein
The quantization parameter conversion means, when the block determination means determines that the sub-block in the macro block is encoded as an intra block, the quantization parameter corresponding to the sub-block as a quantization parameter Converted to parameters,
21. The image encoding apparatus according to claim 20, wherein the entropy encoding means entropy encodes the orthogonal transform mode and includes it in the bitstream.
As a result of entropy decoding of the bit stream, the quantized transform coefficient that is the decoded data of the bit stream, the transition value for transforming the quantization parameter, and whether the sub block in the macro block is encoded as an intra block or as an inter block Entropy decoding means for outputting a result of determination indicating whether the image is encoded, prediction image generation information used when the prediction image is generated, and a quantization parameter corresponding to the macroblock, and output from the entropy decoding means When the determined determination result indicates that the block is an intra block, the quantization parameter corresponding to the macro block is converted into the quantization parameter corresponding to the sub block according to the transition value output from the entropy decoding means. Output the converted quantization parameter, A quantization parameter conversion unit that outputs a quantization parameter corresponding to the macroblock as a quantization parameter corresponding to the sub-block, when the determination result output from the entropy decoding unit indicates an inter block; Inverse quantization that inverse quantizes the transform coefficient output from the entropy decoding means using the quantization parameter corresponding to the sub-block output from the quantization parameter conversion means, and inversely transforms the transform coefficient after the inverse quantization. A prediction image generation unit that generates a prediction image using the prediction image generation information output from the entropy decoding unit, the prediction image generated by the prediction image generation unit, and the inverse quantization unit And a decoded image generating means for generating a decoded image by adding a difference image as an inverse orthogonal transformation result Image decoding device.
For each of the various sub-blocks, conversion table storage means is provided for storing a conversion table that records the correspondence between the quantization parameter corresponding to the macroblock and the quantization parameter corresponding to the sub-block. If
If the quantization parameter conversion means indicates that the determination result output from the entropy decoding means is an intra block, a prediction image is generated from the plurality of conversion tables stored by the conversion table storage means. The quantization parameter corresponding to the macroblock is converted into the quantization parameter corresponding to the subblock by referring to the conversion table related to the subblock used in the intra prediction process performed when the conversion is performed. 26. The image decoding apparatus according to claim 25, wherein a quantization parameter is output.
The entropy decoding means, as decoded data of the bitstream, changes between a quantization parameter corresponding to a macroblock and a quantization parameter corresponding to a subblock used in an intra prediction process performed when a predicted image is generated. Output the value
If the determination result output from the entropy decoding means indicates that the block is an intra block, the quantization parameter conversion means adds the transition value to the quantization parameter corresponding to the macro block and corresponds to the sub block. 26. The image decoding apparatus according to claim 25, wherein a quantization parameter to be calculated is calculated and a quantization parameter corresponding to the sub-block is output.
The entropy decoding means outputs an intra prediction type as decoded data of the bit stream,
If the determination result output from the entropy decoding means indicates an intra block, the quantization parameter conversion means converts the quantization parameter corresponding to the sub-block into a quantization parameter corresponding to the intra prediction type. 26. The image decoding apparatus according to claim 25, wherein:
The entropy decoding means outputs the orthogonal transform mode of the intra block as the decoded data of the bit stream,
If the determination result output from the entropy decoding means indicates that the block is an intra block, the quantization parameter conversion means converts the quantization parameter corresponding to the sub-block into a quantization parameter corresponding to the orthogonal transformation mode. 26. The image decoding apparatus according to claim 25, wherein:
The difference image generation means encodes the input image in units of macroblocks, generates a prediction image from the locally decoded image after encoding, and generates a difference image between the input image and the prediction image A difference image generation processing step, a block determination processing step for determining whether the block determination means encodes a sub-block in the macroblock as an intra block or an inter block, and the block determination processing If it is determined in the step that the sub-block in the macroblock is encoded as an intra block, the quantization parameter conversion means determines the quantization parameter corresponding to the macroblock according to the transition value for quantization parameter conversion. To the quantization parameter corresponding to the sub-block, and the converted quantum A parameter is output, and when it is determined in the block determination processing step that encoding is performed using a sub-block in the macroblock as an inter-block, the quantization parameter conversion means converts the quantization parameter corresponding to the macroblock A quantization parameter conversion processing step that outputs the quantization parameter corresponding to the sub-block, and the quantization means orthogonally transforms the difference image generated in the difference image generation processing step and outputs in the quantization parameter conversion processing step. A quantization processing step for quantizing the transform coefficient of the difference image using a quantization parameter corresponding to the sub-block, and the entropy encoding means, the transform coefficient quantized in the quantization processing step, Transition value, judgment result in the block judgment processing step, difference An entropy encoding processing step for generating a bitstream by entropy encoding the prediction image generation information used when the prediction image is generated in the image generation processing step and the quantization parameter corresponding to the macroblock. Image coding method.
The entropy decoding means entropy decodes the bit stream, and the quantized transform coefficient that is the decoded data of the bit stream, the transition value for quantization parameter conversion, and the sub-block in the macro block are encoded as an intra block. An entropy decoding processing step for outputting a determination result indicating whether the current image is encoded as an interblock, prediction image generation information used when the predicted image is generated, and a quantization parameter corresponding to the macroblock; When the determination result output in the entropy decoding processing step indicates that the block is an intra block, the quantization parameter conversion means determines the quantum corresponding to the macroblock according to the transition value output in the entropy decoding processing step. Parameter to the corresponding sub-block When converting to a corresponding quantization parameter, outputting the converted quantization parameter, and indicating that the determination result output in the entropy decoding processing step is an inter block, the quantization parameter conversion means, A quantization parameter conversion processing step that outputs a quantization parameter corresponding to the macroblock as a quantization parameter corresponding to the subblock, and an inverse quantization means corresponds to the subblock output in the quantization parameter conversion processing step. Using the quantization parameter to perform inverse quantization on the transform coefficient output in the entropy decoding process step and inverse orthogonal transform on the transform coefficient after the inverse quantization, and the decoded image generation means, Prediction using the prediction image generation information output in the entropy decoding process step A predicted image generation processing step that generates an image, and a decoded image generation unit that adds the predicted image generated in the predicted image generation processing step and the difference image that is the result of the inverse orthogonal transform in the inverse quantization processing step. An image decoding method comprising: a decoded image generation processing step for generating a decoded image.