WO1997047140A1

WO1997047140A1 - Device and method for encoding animations data and device and method for decoding animations data

Info

Publication number: WO1997047140A1
Application number: PCT/JP1997/001918
Authority: WO
Inventors: Hiroyuki Akutsu
Original assignee: Dome Inc.
Priority date: 1996-06-07
Filing date: 1997-06-05
Publication date: 1997-12-11
Also published as: AU2979097A

Abstract

Animation data are transmitted in accordance with their characteristics. An encoded reference picture data storing means (105) stores the picture data of one screen constituted of a plurality of picture elements as encoded reference picture data. A differential evaluation computing means (107) computes the difference evaluation between block data constituting picture data to be encoded and block data corresponding to the encoded reference picture data. An encoding and transmitting means (109) encodes block data which are different from the block data corresponding to the encoded reference picture data as block data to be encoded based on the differential evaluation and transmits the encoded block data together with a block ID by changing a transmitting method.

Description

Specification

[Title of Invention]

Moving picture data encoding apparatus, moving picture decoding apparatus and method thereof

[Technical field]

The present invention relates to an apparatus for encoding or decoding moving image data, and more particularly, to specifying an encoding target portion.

[Background technology]

Since moving image data is data in which still images are arranged in chronological order, basically, moving image data can be displayed by sequentially transferring still image data for each frame.

Today, attention has been paid to the MC (or ion on Compensation) method as an encoding method for video data. The MC method is a method in which a key frame is transmitted as it is for one frame, and in the next frame, only data different from the key frame is transferred.

However, the conventional MC method requires a great deal of computation time to determine which object has moved to where. This is because it takes time to determine which range of image is the one object.

[Disclosure of the Invention]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a moving picture data encoding apparatus or method capable of transmitting or encoding according to the characteristics of moving picture data.

In the moving picture data encoding device according to the present invention,

A useful means for recording encoded reference image data in which one screen of image data composed of a plurality of pixels is used as encoding reference image data.

The given encoding target image data is divided into a plurality of block data, and for each of the divided block data, block data different from the corresponding block data of the encoding reference image data is set as the encoding target block data. A coded transmitting means for transmitting the coded block data together with the block ID by coding;

In the moving image data encoding device provided with A differential evaluation operation unit for calculating a difference evaluation between block data constituting the image data to be encoded and corresponding block data of the encoding reference image data, wherein the encoding transmission unit comprises: Changing the transmission method of the given image data to be encoded based on the difference evaluation of

It is characterized by.

In the moving picture data decoding device according to the present invention,

A moving image data decoding device for receiving encoded data transmitted from the moving image data encoding device and decoding the encoded data,

A) Convenient means for storing encoded reference image data, which stores image data for one screen composed of multiple pixels as encoded reference image data.

B) Given coded block data obtained by coding the block data and its block ID, decoding means for decoding the coded block data;

C) a next frame image data generating means for generating image data of a frame next to the reference image data based on the encoded reference image data and the decoded block data;

D) encoding reference image data rewriting means for rewriting the encoding reference image data of the encoding reference image data storage means to the next frame image data;

It is characterized by having.

In the moving image data encoding method according to the present invention,

Image data for one screen composed of a plurality of pixels is kept as a criterion for encoding reference image data, and given image data to be encoded is divided into a plurality of block data, and each divided block is For the data, video data that is coded as block data to be coded, which is different from the block data corresponding to the coded reference image data, and the coded block data of the coded block is transmitted together with its block ID, and is transmitted as moving image data. In the encoding method,

A difference evaluation between the block data constituting the image data to be encoded and the corresponding block data of the encoding reference image data is calculated, and a given code is given based on the difference evaluation of each block data. To change the transmission method of the image data to be converted, It is characterized by.

In the moving picture data decoding method according to the present invention,

A moving image data decoding method for receiving encoded data and decoding the encoded data, wherein an image data for one screen composed of a plurality of pixels is stored as reference image data,

Given the coded block data obtained by coding the block data and its block ID, the block data after this coding is decoded,

Based on the reference image data and the decoded block data, generate image data of a frame next to the reference image data,

Rewriting the encoded reference image data with the next frame image data.

In the moving picture data encoding device according to the present invention,

Encoding reference image data storage means for storing image data of one screen composed of a plurality of pixels as encoding reference image data;

The given encoding target image data is divided into a plurality of block data, and for each of the divided block data, block data different from the corresponding block data of the encoding reference image data is set as encoding target block data. Encoding means for determining and encoding the ID corresponding to the block ID,

In the moving image data encoding device provided with

The image processing apparatus further includes a difference evaluation operation unit configured to calculate a difference evaluation between the block data configuring the image data to be encoded and the corresponding block data of the encoding reference image data. Determining block data to be encoded based on the difference evaluation;

It is characterized by.

In the moving picture decoding apparatus according to the present invention,

A moving image data decoding device for decoding the encoded data encoded by the moving image data encoding device,

A) reference image data storage means for storing image data for one screen composed of a plurality of pixels as reference image data; P 7/01918

B) Given encoded block data obtained by encoding block data and its block ID, decoding means for decoding the encoded block data;

C) based on the reference image data and the decoded block data, a next frame image data generation means for generating an image data of a frame next to the reference image data,

D) encoding reference image data rewriting means for rewriting the encoding reference image data of the encoding base image data storage means to the next frame image data;

It is characterized by having.

In the moving image data encoding method according to the present invention,

The image data of one screen composed of a plurality of pixels is useful as an encoding reference image data, and the given image data to be encoded is divided into a plurality of block data and divided. In the moving image data encoding method, for each block data, a block data that is different from the corresponding block data of the encoding reference image data is encoded as the encoding target block data in association with the block ID. A difference evaluation between the block data constituting the image data of the above and the corresponding block data of the coding reference image data is calculated, and the block data to be coded is determined based on the difference evaluation of each block data. thing,

It is characterized by.

In the moving picture data decoding method according to the present invention,

A moving image data decoding method for decoding encoded data,

One screen of image data composed of a plurality of pixels is clerk as coding reference image data,

Given coded block data obtained by coding block data and its block ID, the coded block data is decoded, and

The encoding reference image data is replaced with the next frame image data.

For the features, other objects, uses, effects, and the like of the present invention, refer to the examples and the drawings. And will become clear.

[Brief description of drawings]

FIG. 1 is a functional block diagram of a moving picture data encoding apparatus 110 according to the present invention. FIG. 2 is a functional block diagram of the encoding means 109a.

FIG. 3 is a functional block diagram of the moving picture data decoding device 200 according to the present invention. FIG. 4 is a functional block diagram of the decoding means 206 of FIG.

FIG. 5 is a diagram showing an example of a hardware configuration of the moving picture data encoding device 110 and the moving picture data decoding apparatus 200 shown in FIG.

FIG. 6 is an overall flowchart of the transmission process.

FIG. 7 is an overall flowchart of the transmission process.

FIG. 8 is a diagram illustrating an example of moving image data.

FIG. 9 is a graph illustrating the moving speed of the object in the moving image data shown in FIG. FIG. 10 is a diagram showing image data to be encoded.

FIG. 11 is a diagram illustrating an example of the difference evaluation of each block.

FIG. 12 is a flowchart of the first encoding process.

FIG. 13 is a diagram showing data of one block of image data to be encoded. FIG. 14 is a diagram showing a data structure in the encoding process.

FIG. 15 is a diagram showing a data structure of a codebook used for encoding and decoding.

FIGS. 16A to 16E are diagrams for explaining the transition of some blocks of input data and their corresponding comparison buffers.

FIG. 17 is a flowchart showing the entire decoding process of the received block data.

FIG. 18 is a flowchart showing details of the decoding process.

FIG. 19 is a flowchart of the encoding process (second embodiment).

FIG. 20 is a flowchart of the decoding process (second embodiment).

FIG. 21 is a diagram for explaining an arrangement relationship of index numbers assigned to each block data in the learning process.

FIG. 22 is a flowchart of the learning process. FIG. 23 is a diagram showing a Hadamard transform matrix.

[Best mode for carrying out the invention]

"table of contents"

1. Overall block diagram

1) Video data encoding device 1 1 0

1) Overall function block diagram

2) Encoding means 1 0 9 a

(2) Video data decoding device 2 0 0

1-2-1) Overall functional block diagram

(2) 2) About decryption means 206

2. Hardware configuration

3. Flow chart

3-1) Transmission processing

3- 2) Reception processing

4. Other data encoding and decoding methods

4- 1) Second encoding method

4-U) Function block diagram

4) 2) Flow chart

5. About creating a codebook

6. Other Embodiments One embodiment of the present invention will be described with reference to the drawings.

1. Overall block diagram

1) Video data encoding device 1

1-1) Overall functional block diagram

The moving image data encoding device 110 shown in FIG. 1 includes encoding target image data storage means 103, encoding reference image data useful means 105, difference evaluation operation means 107, encoding transmission Means 109, and means for rewriting encoded reference image data 108.

Encoding target image data storage means 1 0 3, _c to Ki慷images de Isseki encoding target The encoding reference image data storage means 105 stores image data for one screen composed of a plurality of pixels as encoding reference image data.

The difference evaluation calculating means 107 calculates a difference evaluation between the block data constituting the image data to be encoded and the corresponding block data of the encoded reference image data.

The encoding transmission means 109 divides the given image data to be encoded into a plurality of block data, and, for each of the divided block data, a block data different from the corresponding block data of the above-mentioned encoded reference image data. The encoded data is encoded as the encoding target block data, and the encoded encoded block data is transmitted together with the block ID. Note that the coded transmission means 109 changes the transmission method for the given image data to be coded based on the difference evaluation between each block of data, and transmits the coded data. I do.

The coding reference image data rewriting means 108 rewrites the block reference of the coding reference screen data with the block data determined as the coding target. The encoding transmission means 109 changes the transmission method according to the evaluation value distribution of each difference evaluation of the block data of the image data for one screen. Changes in the transmission method include changes in pre-transmission processing and transmission conditions.

The coded transmission unit 109 detects the transmission state, and changes the transmission method based on the difference evaluation and the detection result.

Also, the coded transmission means 109 can change the transmission method by comparing with a preset difference threshold value.

Further, the encoding transmission unit 109 extracts transmission target block data from a plurality of encoding target block data regardless of the evaluation value distribution of each difference evaluation of the block data of the image data for one screen. I do.

2) Encoding means 1 0 9 a

The means for performing the coding process of the coding transmission means 109 (hereinafter referred to as coding means 109a) will be described with reference to FIG. The encoding means 109a stores a plurality of block data composed of a plurality of block data and a plurality of codes corresponding to the data as candidate data. Given, Encoding is performed using the plurality of candidate data. The encoding transmission unit 109 a includes a relative block data storage unit 9, a first conversion unit 3, a relative block data encoding unit 7, and a block characteristic data calculation unit 5.

The relative block data useful means 9 uses the relative block data and a code corresponding to the relative block data as candidate data. In the present embodiment, as the candidate data, a plurality of group correspondence tables, which are grouped according to the block characteristic data, are used.

The first conversion means 3 determines a conversion reference value in the block data based on the block data to be encoded, and expresses each block configuration data constituting the block data as a relative value with respect to the conversion reference value. Is converted to relative block data. In the present embodiment, the minimum value of each block configuration data configuring the block data to be encoded is determined as the conversion reference value in the block data.

The block characteristic data calculating means 5 calculates block characteristic data of the block data based on the distribution characteristic of each block configuration data constituting the block data for the block data to be encoded. In the present embodiment, as the distribution characteristic, difference data between the maximum value and the minimum value of each block configuration data configuring the block data to be encoded is referred to as block characteristic data in the block data. Asked.

The relative block data encoding means 7 encodes the relative block data using the candidate data stored in the relative block data storage means 9, and stores the relative code data and the conversion criterion. Output the value. In the present embodiment, based on the block characteristic data, which group of candidate data stored in the relative block data useful means 9 is used as the relative code data. And adopted this group name and intra-group number.

(2) Video data decoding device 2 0 0

2-1) Overall functional block diagram

The moving image data decoding device 200 shown in FIG. 3 is a device that receives the encoded data transmitted from the moving image data encoding device 110 and decodes it. Video data recovery The encoding device 200 includes an encoding reference image data storage unit 205, a decoding unit 206, a next frame image data generation unit 207, and an encoding reference image data rewriting unit 209. I have.

The coded reference image data storage means 205 stores one screen of image data composed of a plurality of pixels as the coded reference image data.

Decoding means 206, given the coded block data obtained by coding the block data and the block ID thereof, decodes the coded block data.

The next frame image data generation means 207 generates image data of a frame next to the reference image data based on the encoded reference image data and the decoded block data. The encoding reference image data rewriting means 209 rewrites the encoding reference image data in the encoding reference image data storage means to the next frame image data.

In this embodiment, the conversion reference value is determined for each block for each encoding target block data. However, one conversion reference value may be determined for each of the predetermined number of blocks. As a result, the amount of data after compression can be reduced. This is particularly effective when the block data is correlated as in image data.

1 -2- 2) Decoding means 206

Details of the decoding means 206 will be described with reference to FIG. The decoding means 206 shown in FIG. 4 is an apparatus for decoding the coded data coded by the coding means 109a shown in FIG. The decoding means 206 receives the relative block data which is obtained by coding the relative block data converted by using the block data conversion reference value composed of a plurality of block configuration data, and the conversion reference value of the block data is If given, decrypt it. The decoding means 206 includes a relative block data recording means 9, a second conversion means 13, and a relative block data decoding means 17. The relative block data useful means 9 stores a plurality of relative block data and codes corresponding thereto as candidate data. In the present embodiment, a group correspondence table, which is grouped according to block characteristic data, is used as candidate data. Multiple useful.

The relative block data decoding unit 17 decodes the relative code data using the candidate data stored in the relative block data storage unit. Specifically, when a group name and an intra-group number are given as the relative code data, the group name and the intra-group number determine which candidate data of the group correspondence table is to be used, Decoding is performed in the corresponding relative block overnight. The second conversion means 13 converts the decoded relative block data into relative block data before encoding using the conversion reference value in the block data. In the present embodiment, a description has been given of a case where a conversion reference value is added for each block for each block data to be coded, but a block before coding is converted using a conversion reference value for each of a predetermined number of blocks. You may make it convert into data.

2. Hardware configuration

FIG. 5 shows an example of a hardware configuration in which the moving picture data encoding device 110 shown in FIG. 1 is realized using a CPU.

The video data encoding device 21 includes a CPU 23, a memory 27, a hard disk 26, a CRT 30, a FDD 25, a keyboard 28, a transmission control unit 32, and a pass line 29.

The CPU 23 controls each unit via a bus line 29 according to a control program stored in the hard disk 26.

The control program is read from the flexible disk to which the program is stored via the FDD 25 and installed on the hard disk 26. In addition to the flexible disk, a hard disk may be installed from a computer-readable pathological medium in which a program such as a CD-ROM or an IC card is physically integrated. Further, a down port may be provided using a communication line.

In this embodiment, by installing the program from the flexible disk to the hard disk 26, the program useful for the flexible disk is indirectly executed by the computer. However, the present invention is not limited to this. It may be executed directly from 5. Computer-executable programs include those that can be directly executed by simply installing the software as it is, as well as those that need to be converted to another form (for example, decompress data-compressed programs). Etc.), and those that can be executed in combination with other module parts.

The hard disk 26 stores a code book described later. The memory 27 has a comparison buffer 26a, a frame buffer 26b, and a block buffer 27c. The comparison buffer 26a stores an encoded reference image described later. The frame buffer 26 b stores the image data for one screen to be encoded. The block buffer 27c stores encoded block data. The operation result and the like are temporarily stored in the memory 27. In CRT30, moving image data before encoding and the like are displayed. The transmission control unit 32 transmits and receives data by wire or wirelessly. The keyboard 28 is an instruction input means for inputting a read instruction or the like.

Note that the hardware configuration of the moving image data decoding device 200 is almost the same as that of the moving image data encoding device 110. The block buffer 27c stores the received encoded block data. The comparison buffer 26a is unnecessary. The frame buffer 26 b stores the decoded image data for one screen. The frame buffer 26b is sequentially rewritten with the sequentially decoded block data as described later.

3. Flow chart

3-1) Transmission processing

A flowchart for determining the block data to be encoded by the moving picture data encoding apparatus 110 will be described with reference to FIGS.

In the following, as shown in FIG. 8, the star mark 160 indicates the time t. Between the time t _4, moves in the arrow α direction by a distance L 1, description will be given of a case where a round mark 1 6 2 is moved by a distance L 2 in the arrow α in the same time. The relationship between the movement of the star mark 160 and the round mark 16 2 will be described with reference to FIG. The star mark 160 is at time t, as shown by line 72. Moved by a distance L 1 between times t ₃ from between the time of the time t ₄ is stopped. On the other hand, the circle mark 16 2 indicates that 1 I-

Replacement paper (26) Time t. Between the t _4, as compared with the distance L 1 has moved a short distance L 2. That is, the star mark 160 moves faster than the round mark 162.

First, the CPU 23 initializes the comparison buffer 27a (step ST101 in FIG. 6), and determines whether or not encoding target image data has been input (step ST103). If the encoding target image data is not input, the process ends.

On the other hand, when the encoding target image data is input via the transmission control unit 31, the CPU 23 stores the encoding target image data in the frame buffer 27b of the memory 27 (step ST105).

Next, the CPU 23 determines whether or not it is the first image (first frame) (step ST107). In this case, since it is the first image, the process proceeds to step ST130 in FIG. 7 to divide the data into block data. In the present embodiment, as shown in FIG. 10, image data for one screen is divided into block data constituting 16 pixels of 4 <4.

Next, the block data is encoded (step ST131). In this case, since this is the first image, encoding is performed for the entire block data. The encoding process will be described with reference to FIG.

First, the CPU 23 initializes the processing target block number h (ST3 in FIG. 12), and among the image data for one screen stored in the frame buffer 26b of the hard disk 26, reads the block data of the hth block. Evening is read from memory 27 (step ST5).

For example, if the block configuration data constituting the block data Aij of the ij-th block is aij (0) to aij (15) as shown in FIG. 13, the block data of the ij-th block is represented by Aij = (aij (0>, a ij (1), ..., aij (15)), the block data A00 of the 00th block is composed of the block configuration data a00 (0) to a00 ( According to 15), it is expressed as follows.

A00 = (a 00 (0), a 00 (1),

In the present embodiment, the block configuration data aij (0) to aij (15) constituting each block take any value from 0 to 255. That is, the values of the block configuration data a ij (0) to a U (15) are represented by 8 bits.

I 2 ― Notation is given in decimal notation for easy understanding.

Here, the description will be made assuming that the block data A00 of the 00th block is represented by the block configuration data shown in FIG. 14A.

The CPU 23 obtains shift block configuration data AO 0 'obtained by subjecting each block configuration data a 00 (0) to a00 (15) of the block data A00 shown in FIG. Two-step ST 1 1). In this case, the shift block configuration data A00 'is represented as shown in Fig. 14B.

This shift block configuration data A00 'is expressed as AO0' = (aOO (OK, aOO (l)) • ··· a00 (15) « ⁴ ) In this case, the block configuration data a00 (0) ~ Since a00 (15) is 8-bit data, shift block configuration data aOO (OK to a00 (15) <G is 4-bit data. That is, shift block configuration data a00 (0) < Each of the groups a00 (15) has a value of 16. For example, if the block configuration data aij (0) is “0 1 00 1 1 1 0”, the shift block configuration data a00 (0 ) <Gha becomes a 4-bit data of “0 1 00” (quantization).

Next, the CPU 23 obtains the maximum value maxAOO 'and the minimum value minA00' among the shift block configuration data a00 (0) through a00 (15) (FIG. 12, step ST13). In this case, the maximum value maxAOO '= 16 and the minimum value minA00' = 2 are obtained for the block data A00 '(see FIG. 14B) and stored in the memory 27. The maximum value maxAOO 'and the minimum value minA00' are also 4-bit data. Next, the CPU 23 obtains a difference data between the maximum value maxAOO 'and the minimum value minAOO'—evening AOOg (FIG. 12, step ST15). Specifically, the difference data A00 g is obtained by the following equation.

A00g = max x A00 'one min AOO' = 1 6-2 = 14

The difference data A 00 g indicates how much the maximum value and the minimum value are different for each block configuration data of the block data A 00. That is, when the difference data AOOg is large, the difference between the maximum value and the minimum value is large, and it can be determined that the block data A00 is an image having an edge. On the other hand, an image having a small difference between the maximum value and the minimum value can be determined to be an image that is not so. Next, the CPU 23 shifts the difference data AOOg by one bit to the right, and shifts the right one bit by one bit.

Replacement form (Rule 26) Find shift difference data AOOg '. Since the difference data AOOg is represented by 4 pits, the right 1-bit shift difference data AOOg 'is represented by 3 bits. In this case, one-bit shift difference AOOg '= 7.

Next, the CPU 23 determines the relative reference value of the block, and subtracts the relative reference value from each of the block configuration data aOO (OK to ai (15)) constituting the shift block data AO0 ', Find block configuration relative data

(FIG. 12, step ST 19). In this embodiment, the minimum value minAOO 'is set as the relative reference value.

If this block configuration relative data is expressed as b00 (0K g ~ b00 (15) <g, the block configuration relative data b00 (0) guru ~ bOO (15K g is expressed by the following formula. .

b00 (0) << ⁴ = a00 (0> 〈guichi mi η ΑΟΟ '

bOO (l) ku = aOO (l) ku—min i ΑΟΟ '

b00 (15) « ⁴ = a00 (15)« ⁴ -mi nAOO '

Note that the maximum value maxAOO 'may be used as the relative reference value. As described above, by using the minimum value or the maximum value as the relative reference value, a later-described calculation of a deviation value becomes easy. The relative reference value is not limited to this, and may be any one of the values in the block configuration data, and may be an arbitrary value.

In this way, for the shift block configuration data ai (0) <g ~ ai (15K, the minimum value min nOOO 'is set to a relative value of 0. If this is the block relative data- relative de Isseki B0O is, _c represented as shown in FIG. 14 C

B00 = (b0O (0K, b00 (l) cug, · · · ~, b00 (15) cug)

Next, the total deviation TOij is calculated. The total shift value TO00 of the index number 00 can be obtained by the following equation.

TO00 = (ABS (b00 (0K × 1 k00 (0))) + ABS (b 00 (1) « ⁴ -k 00 (1)) + ...- + ABS (b00 (15)« ⁴ -k00 (15) )}

Note that ABS 0 represents an operation for obtaining an absolute value within 0. This is performed up to the index number 255 of the group number “7 J of code book K” shown in FIG.

For example, when the total deviation value TO 245 from the relative block data K245 is the minimum, the block configuration relative data B00 is encoded into the group number “7” and the index number “245”. This is called a relative sign.

Next, the CPU 23 stores the relative encoded data and the minimum value minAOO 'in the block in the block buffer 27c as compressed data, and ends the encoding of the 00th block (step ST23 in FIG. 12).

After all, the block data AO0 is, after all, the group number “7” indicated by the shift difference data A i g ′, the index number “245” of the group and the minimum value m

1 nAOO '“This means that the data has been compressed to 2J. The group number is 3 bits, the index number is 8 bits, and the minimum value is 4 bits, so the data is compressed to a total of 15 bits. Thus, one block of data is compressed into 15-bit data.

Next, in step ST25 in FIG. 12, the CPU 23 determines whether or not encoding has been completed for all blocks, and if not completed, increments the processing target block number h (see FIG. 12). Step ST26), the processing of steps ST5 to ST25 is repeated.

In step ST25 in FIG. 12, when the encoding is completed for all the blocks, the encoding of the block data is terminated.

When encoding is completed for all block data, the CPU 23 compares the image data of the first frame stored in the frame buffer 27 b with a comparison buffer.

27a is overwritten (step ST133 in Fig. 7). As a result, the image data of the first frame is stored in the comparison buffer 27a. Further, the CPU 23 transmits the encoded block data stored in the block buffer 27c together with the corresponding block ID via the transmission control unit 31 (step ST127). Specifically, for each block, the relative sign data and the minimum value are transmitted together with the block ID. In this way, the image data of the first frame is P TJP97 / 01918

Is transmitted in a compressed state.

Next, a case where the image data of the second frame is provided will be described. The CPU 23 determines whether or not encoding target image data has been input (step ST103 in FIG. 6). When the image data of the second frame is input, the image data of the frame is usefully stored in the frame buffer 27b (step ST105). In this case, since it is not the first plane image, the process proceeds to step ST109, and the image data stored in the frame buffer 27b is divided into block data (step S109). This division processing is the same as step ST130 in FIG. 7, and thus description thereof is omitted.

Next, the CPU 23 initializes the processing target block number h (step ST111 in FIG. 6). Then, the CPU 23 reads the h-th block data, and obtains a difference evaluation between the h-th block data of the comparison buffer 26a and the h-th block data of the frame buffer 26b (step ST113). .

In this case, since the block number to be processed is h = 0, the image data of the first frame stored in the comparison buffer 27a corresponding to the 0th block data corresponds to the frame buffer 26b. Calculate difference evaluation with block data.

This difference evaluation is performed by comparing the acquaintance of each corresponding pixel. Specifically, first, for the 0th pixel of the 0th block data, the absolute value of the difference data is obtained. For example, the density of the 0th pixel of the 0th block data used in the comparison buffer 26a is “D 0a”, and the chromaticity of the 0th pixel of the 0th block data stored in the frame buffer 26b is "If D0bj, the absolute value of the difference data for the 0th pixel of the 0th block data is expressed as ABS (DOa-DOb). Note that ABS0 is the absolute value within 0. In the same way, calculate the absolute value of the difference data of the first pixel, calculate it for 16 pixels, and sum up all the values.This total value is the difference evaluation of the block data.

Next, the CPU 23 determines whether or not the calculation of the difference evaluation has been completed for all the block data (step ST115). In this case, the 0th block data —Since the calculation of the evaluation value has been performed only in the evening, the processing target block number h is incremented (step ST117), and the processing of step ST113 to step ST115 is repeated. If it is determined in step ST115 that the difference evaluation of all block data has been obtained, the process proceeds to step ST121 in FIG. 7, and the block number to be encoded is determined. For example, block A as shown in Figure 11. . Has a rating of “0”, block A «has a rating of“ 0 ”, block A. _If the evaluation of ₂ is “10” and the evaluation of block Ao ₃ is “20”, the block number to be coded is determined as follows.

The CPU 23 sequentially extracts a predetermined number of blocks having a large evaluation value from the block data. In this embodiment, 50 blocks are selected from the blocks having the highest evaluation values among all the blocks. If the block number A 03, block number A ₁₃ and block number A ₁₃ shown in FIG. 11 are the largest, and the block number A _{12 is} the next largest, the block data of 50 Determined as four block data.

CPU 23 performs an encoding process on the block data determined to be encoded (step ST123). Such an encoding process is the same as that described in step ST131, and a description thereof will be omitted.

Next, the CPU 23 updates the image data stored in the comparison buffer 27a only for the pixels corresponding to the block data determined to be encoded in step ST121 (step ST125). ). In this case, since 50 blocks of the entire block data are to be encoded, only the pixels corresponding to the 50 blocks of the entire block data are image data stored in the comparison buffer 27a. Be updated.

Next, the CPU 23 transmits only the encoded block data (step ST127). In this case, the coded block data is only the block data to be coded in step ST 121, and therefore, unlike the first frame, the second frame corresponds to a different block. Only pixel data is compressed and transmitted together with its block ID.

The same applies to the third and fourth frames' 1 1

Only the image data corresponding to the block different from the image in the buffer 26a is data-compressed and transmitted together with the block ID. Then, only the pixels corresponding to the pixels belonging to the transmitted block are updated in the comparison buffer 26a for the next frame.

As described above, in the present embodiment, not all block data different from the first frame is transmitted, but data encoding and data transfer are performed only for blocks having a high evaluation value but within a certain range. Therefore, as shown in Fig. 8, when the star mark 160 moves fast and the circle mark 162 moves slowly, the star mark 160 is transmitted in each frame. On the other hand, when the circle mark 162 moves for a certain distance, the data is transmitted only when the difference evaluation value in each block data becomes large. That is, for such a slow-moving object, there is a high possibility that the evaluation value becomes small, so that the transfer is performed only when the moving amount becomes large to some extent.

This will be described specifically with reference to FIG. FIG. 16 shows the time t ₃ in FIG. 9 for a part of the block data 71 of the star mark 160 and a part of the block data 73 of the circle mark 16 2 shown in FIG. The transition from immediately before to immediately after is shown.

First, a description is given to the block data 7 1 star mark 1 6 0 immediately before time t ₃ in FIG. In this case, the block de night 71 has eight pixels on the left side as shown in FIG. 16A. Next, in the next frame, you can see that it has moved two pixels to the right. In the next frame as well, it is similarly moved two pixels to the right. Then, in the next frame, it stops in that state as shown in Fig. 16D. Furthermore, it stops at the next frame as shown in Figure 16E. On the other hand, as for some blocks of the circle mark 16 2 and the block 73, as shown in FIG. 16A, eight pixels in the middle two columns of the 16 pixels are darkened. Then, in the next frame, there is no movement (see Fig. 16B), and in the next third frame, it moves rightward by one pixel (see Fig. 16C). Similarly, in the next frame, there is no movement (see Fig. 16D), and in the next frame, it moves one pixel in the direction (Fig. 16E).

― I 8 ― Thus, the block 71 moves to the right by two pixels for each frame, and stops in the state shown in FIG. 16C. On the other hand, the block 73 moves to the right one for the first time every two frames. In such a situation, as described below, the block 71 is selected as the block to be encoded, while the block 73 is selected as the block to be encoded in relation to the corresponding comparison buffer. It will not be selected as a block.

First, consider the case where the data content of the block 71 changes from the state shown in FIG. 16A as shown in FIG. 16B. The block 71 shown in Fig. 16B differs from the corresponding comparison buffer (see Fig. 16A) in that the density of the eight pixels in the right two columns differs in this case. The difference between the two pixels is the difference evaluation value in block 71. On the other hand, for the block 73, since the data shown in FIG. 16B and the contents of the corresponding comparison buffer shown in FIG. 16A are the same, the difference evaluation value is zero (0). Here, if block 71 is determined to be a block to be encoded, the contents of the corresponding comparison buffer are also overwritten with the data of block 71 in FIG. 16B, as shown in FIG. 16B. Is made.

Next, when the data changes from the state shown in Fig. 16B to the state shown in Fig. 16C, similarly for block 71, the difference in the density of eight pixels in the middle two columns is the difference evaluation value and the difference evaluation value. Become. On the other hand, for the block 73, the difference in density between the four pixels in the rightmost column in total is the difference evaluation value. In this case, similarly, for the block 71, a difference of eight pixels is obtained as an evaluation value. On the other hand, for the block 73, a difference of four pixels in one column on the right side is determined as an evaluation value. Therefore, as described above, if a block having a large evaluation value is selected, the block 73 may be excluded from coding. If excluded, the contents of the corresponding comparison buffer in block 73 will not be rewritten (see Figure 16C).

Next, a case in which the state shown in FIG. 16C is shifted to the state shown in FIG. 16D will be described. In this case, since the star mark 160 is stopped for the block 71, there is no difference between the contents of the block 71 shown in FIG. 16D and the corresponding comparison buffer shown in FIG. 16C. Therefore, the difference evaluation value is zero (0). On the other hand, regarding the block 73, the state shown in FIG. 16D and the state shown in FIG.

I 9- The data for four pixels on one side is different. In the previous section, since block 71 had an evaluation value with a difference value of 8 pixels, the difference of 4 pixels was excluded from the block to be encoded without exceeding the threshold value. Was. However, in this state, the star mark 160 shown in FIG. 8 is stopped, so that even the data of the block 73 having a small difference evaluation value is selected as the encoding target block. As a result, the contents of the corresponding comparison buffer are rewritten as shown in FIG. 16D.

For such an object that changes slowly, transmission is performed such that the object moves as an image for the first time with a delay of several frames.

Next, when moving from the state shown in FIG. 16D to the state shown in FIG. 16E, since the block 71 has already stopped, the difference evaluation value is also zero (0). On the other hand, the block 73 is further moved to the right by one pixel. Also in this case, in the relationship between the contents of the corresponding comparison buffer shown in FIG. 16D and the input data shown in FIG. 16E, a difference of four pixels is the difference evaluation value. As before, the block 73 is selected as a block to be encoded and the contents of the corresponding comparison buffer are rewritten.

In this way, for those moving fast, the evaluation value of the block data increases. On the other hand, for those that move slowly, the evaluation value decreases because the data does not move for several frames and the same data continues. Therefore, if a fast-moving object and a slow-moving object are on one screen, when the fast-moving object stops, it is transmitted as a block to be encoded.

Even if the fast-moving and slow-moving objects are on one screen, and the fast-moving one does not stop, the block corresponding to the slow-moving object is transmitted as follows. Sometimes,

In the present embodiment, in the encoded reference image data of the comparison buffer 26a, only the pixels transmitted after being encoded are updated, and all the pixel data of one frame written in the frame buffer 26b is transmitted. It is not done. In other words, the comparison buffer 26a does not preserve the image data of the previous frame, but rather uses the image data that has been transmitted as a result. In this situation, the image data is moving slowly and only slowly.

— 2 »— Even if there is another object with a high moving speed, the comparison buffer 26a cannot be replaced with a portion having a low moving speed. However, the difference evaluation is obtained by comparing the data of the comparison buffer 26a with the data of the frame buffer 26b. As described above, in consideration of the fact that the comparison buffer 26a is not rewritten, even if the moving speed is slow, it may be determined as a block to be encoded.

As a result, of the moving image data, the motion that humans recognize well, that is, the part that moves greatly and quickly, is transmitted. In addition, even if a slow moving object is present on the same screen as a fast moving object, if it moves to a certain extent, transmission processing is performed as a moving object.

3-2) Reception processing

Next, a flowchart in the case where the moving image data decoding device 200 receives the data will be described with reference to FIG. In this case, it is assumed that the data of each frame shown in FIG. 8 is sent.

In the video data decoding apparatus 200, the memory 27 is initialized when the power is on, and the block data can be received (reception standby state).

The CPU 23 of the video data decoding device 200 determines whether or not the block data after encoding is stored in the block buffer 27c (step ST144), and ends if it is not stored. I do.

If the coded block data transmitted from the transmitting side is useful for the block buffer 27c, the block number h to be processed is initialized (step ST144), and the decoding of the h-th block is performed. (STEP ST149).

The decoding process according to the present embodiment will be described with reference to FIG.

The CPU 23 reads the compressed data of the h-th block from the hard disk 26 and stores it in the memory 27 (step ST33 in FIG. 18). In this case, since the block number to be processed is h = 00, the compressed data of the 00th block is read. As described in the processing of the moving image data encoding device 110, the compressed data of the 00th block includes the group number “7”, the index number “2 4 5” of the group, and the It consists of the small value minAOO '"2".

The CPU 23 uses the codebook K stored on the hard disk 26 to decode the compressed data stored in the memory 27. In this case, the compressed data of the 00th block is composed of a group number “7”, an index number “245” of the group, and a minimum value min A00 ′ “2”. Therefore, it is decoded into relative block data specified by the group number “7” and the index number “245”.

Next, the CPU 23 converts the relative block data using the minimum value minA00 'which is the relative reference value (step ST37 in FIG. 18). More specifically, the minimum value m in ΑΟΟ 'may be added to the relative block configuration data constituting the decoded relative block data.

Next, the CPU 23 shifts the relative block configuration data to which the minimum value minAOO 'has been added by 4 bits to the left (step ST38). As a result, the same 8-bit block configuration data as the block data before encoding is obtained, and the decoding of the 00th block is completed.

In step ST39 in FIG. 18, the CPU 23 determines whether or not decoding has been completed for all blocks. If not, the CPU 23 increments the processing target block number h (step ST40). Steps ST33 to ST38 are returned.

If decoding has been completed for all blocks in step ST39, the decoding ends.

As described above, after conversion into the relative value from the relative reference value, by using the codebook grouped according to the difference data and performing the calculation, the candidate data useful for the codebook is reduced even if it is reduced. However, conversion can be performed without significantly reducing accuracy. Therefore, high accuracy and high-speed conversion are possible, and encoding with a high compression ratio is possible.

In this embodiment, the relative block data Bij whose relative reference value is set to relative 0 is obtained after obtaining the difference data. However, the present invention is not limited to this. Conversely, the difference data may be obtained after obtaining the relative block data Bij. In addition, in this embodiment, when encoding, the total of the absolute values of the shift values is obtained, and the one with the minimum total value is selected, but the conventional method, for example, the following equation As in the above, the sum of the distances may be obtained as the sum of the deviation values.

TO00 = {(b00 (0) ku— k00 (0)) ² + (b 00 (1) « ⁴ -k 00 (I)) ² +

(b00 (15) « ⁴ -k00 (15))" ^{1 2}

In this embodiment, a 4-bit shift process or the like is performed, but such a process may not be performed. In addition, it may be shifted by less than 4 bits or 5 bits or more.

Next, the CPU 23 updates the data in the frame buffer 27b with the decoded data (step ST151 in FIG. 17). The CPU 23 determines whether or not decoding has been completed for all blocks (step ST153). If not completed, the CPU 23 increments the processing target number h, and proceeds from step ST149 to step ST151. Is repeated.

When decoding has been completed for all blocks in step ST153, the CPU 23 outputs the data of the frame buffer 27b to the CRT 30.

(Figure 17 ST 157). In this way, for the first frame, since the data of all the blocks has been compressed and transmitted, the data is decoded, and the decoded data is stored in the frame buffer. Output to

Next, the frame will be described second. For the second and subsequent frames, only blocks that are different from the first frame and that have a high difference evaluation are transmitted. Therefore, it is necessary to combine with the image of the frame immediately before the frame. This synthesis is performed as follows. The data stored in the block buffer 27 is decoded by the decoding method described above. Then, as described in FIG. 17, step ST151, the frame buffer data is updated. In the first frame, the transmitted coded data was all blocks, so all blocks were updated. However, in the following frames, only the blocks that are different from the previous frame and have a high difference evaluation Is transmitted, only the data of the frame buffer corresponding to the block indicated by the block ID is updated (step ST151). For the portion that is not updated, the contents of the frame buffer 26b are output to the CRT 30 as they are.

As described above, in the following frames, moving image data can be displayed by updating only a part different from the previous frame and outputting the updated frame to the CRT 30. Thereby, the moving image data encoded by the encoding device 110 can be received and reliably decoded.

[B: Second Embodiment]

In the first embodiment, vector quantization is performed by quantization and relativization. However, the present invention is not limited to this, and vector quantization can be performed by performing relativization and quantization. Further, after relativization, a left shift operation is performed until the most significant bit of the largest relative block configuration data of the block becomes 1 (hereinafter referred to as a reversible left shift operation). 0 may be added afterwards so that the bit length becomes the same. That is, the shift amount in the reversible left shift operation of each block configuration data is dynamically varied.

The processing in this case will be described with reference to FIG. In this case, it is assumed that each block configuration data e 00 (0) to e 00 (15>) of the block data E00 shown in FIG. 14D is given.

The CPU 23 obtains the maximum value max E00 and the minimum value min E00 from the block configuration data e 00 (0) to e 00 (15) (FIG. 19, step ST 251). In this case, the block data {0} The value maχΕ00 = 48 and the minimum value minE00 = 29 are obtained (see FIG. 14) and stored in the memory 27.

Next, the CPU 23 obtains difference data E00g between the maximum value maXE00 and the minimum value minE00 (step ST253 in FIG. 19). Specifically, the difference data E 00 g is obtained by the following equation.

E00g = m a x E00-min i E00 = 44-29 = 15

Next, the CPU 23 performs a shift operation so that the difference data E00g becomes 3-bit data to obtain shift difference data E00g '(step ST255). As a result, similarly to the above embodiment, a group to be encoded by the codebook K shown in FIG. 15 is specified.

Next, the CPU 23 determines the relative reference value in the block, and subtracts the relative reference value from each of the block configuration data E00 (0) to E00 (15) constituting the block data E0O, Block configuration relative data is obtained (step ST257). In the present embodiment, the minimum value minEOO was used as the relative reference value.

If this block configuration relative data is represented by ί00 (0)-: f00 (15), the block configuration relative data f00 (0) -f00 (15) is represented by the following equation.

f 00 (0) = E00 (0) -ni i nEOO

f 00 (l) = E00 (l) -m i nEOO

f 00 (15) = E00 (15) -min i E0O

Note that the maximum value maxEOO may be used as the relative reference value. As described above, by using the minimum value or the maximum value as the relative reference value, it is easy to calculate a shift value described later. Note that the relative reference value is not limited to this, and may be any one of the block configuration data, and may be an arbitrary value.

In this way, for the block configuration data EOO (0) to E00 (15>), a value is obtained in which the minimum value minEOO is set to relative 0. If this is set as the block relative data F00, the block relative data F00 Is represented as shown in Figure 14E.

F00 = (f 00 (0), f 00 (1), ····, f 00 (15))

Next, the CPU 23 indicates, for the block configuration data having the maximum value among the block configuration data f 00 (0) to f 00 (15), how many bits from the most significant bit MSB continue to 0. The shiftable bit weight Sf00 is calculated (step ST259). In this case, the maximum value of the block configuration data f 00 (0) ~: f 00 (15) is “1 5”, which is expressed in binary notation as “0000 1 1 1 OJ. The shiftable bit amount S f 00 = 4. Since the maximum value of the shiftable bit amount S f 00 is “7”, it is 3-bit data. The CPU 23 performs a reversible left shift operation on the block configuration data ί00 (0) to f00 (15) based on the shiftable bit amount Sί00 (step ST261). For example, since the shiftable bit amount S f00 of the block configuration data f 00 (0) “00001 1 10” is 4, the shift operation is performed to “11 100000”.

The CPU 23 encodes the block relative data F00 'specified by the reversible left-shifted block configuration data f00 (0)' to: fOO (15) '(step ST263). The encoding process is the same as in step ST21 in FIG. 12, and a description thereof will not be repeated.

The CPU 23 stores the relative code data obtained by the encoding and the minimum value minEOO0 in the block as compressed data on the hard disk 26, and ends the encoding of the 00th block (step ST265).

As a result, the block data E0O is data-compressed into the group number indicated by the shift difference data EOOg ', the index number of the group, the minimum value minEOO, and the shiftable bit amount Sf00. Since the group number is 3 bits, the index number is 8 bits, the minimum value is 5 bits, and the shiftable bit SSf00 is 3 bits, 1 block of data (8 bits * 16 = 256 pits) is 19 bits. Compressed overnight.

In this way, after relativization, the shift operation to the left is performed until the most significant bit of the largest relative block configuration data of the block becomes 1 (reversible left shift operation). Can be performed. This is because low-frequency components can be removed by relativizing each block configuration data with the minimum value. This can reduce the number of codebooks. Furthermore, the number of codebooks can be reduced due to the dynamically changing reversible left shift operation amount. That is, if the number of codebooks is the same, accuracy can be improved. Next, decoding will be described.

The CPU 23 initializes the processing target block number h (step ST271 in FIG. 20), shoots out the compressed data of the h-th block from the hard disk 26, and stores it in the memory 27 (step ST273). In this case, since the processing target block number h == 00, the compressed data of the 00th block is read. Compression of block 00 As described above, the data includes a group number, an index number of the group, a minimum value min E00, and a shiftable bit amount S f00. Next, the CPU 23 decodes the compressed data read into the memory 27 using the codebook stored in the hard disk 26. Specifically, CPU 23 decodes the data into relative block data specified by the group number and the index number (step ST275).

Next, the CPU 23 performs a reversible right shift operation using the shiftable bit amount S f00 (step ST277). For example, the block configuration data f 00 (0) ′ “1 1 10 O 000J becomes“ 0000 1 1 10 ”.

CPU 23 converts the relative block data using minimum value minEOO, which is a relative reference value (step ST 278). More specifically, the minimum value minE00 is added to the relative block configuration data constituting the decoded relative block data. Thus, decoding of the 00th block is completed.

In the subsequent processing, as in step ST153 of FIG. 17, the processing of step ST273 to step ST278 is repeated for all blocks until decoding is completed.

In this embodiment, even if the bit length of the shiftable bit amount S f ij is variable, the bit lengths of the block configuration data f 00 (0) ′ to f 00 (15) ′ subjected to the reversible left shift operation are Does not change. This makes it possible to reduce the effective number of significant digits by the amount of shiftable bits S f ij and to simplify the arithmetic processing. However, the present invention is not limited to this, and the bit length of the block configuration data {00 (0) ′ to f00 (15) ′ may be variable. For example, if the shiftable bit amount S f00 of the block number 0 is 4 and the block configuration data f 00 (0) “000001 10” of the block number 0, the reversible shift block shifted left 4 bits The configuration data f 00 (0), is “0

1 ro i i oj instead of 100000 j.

In the first and second embodiments, the relative reference value is obtained for each block. However, one relative reference value may be set for a plurality of blocks. As a result, the same relative reference value can be used in a plurality of blocks, so that the compression ratio can be increased accordingly. For example, 2 * 2 = 4 blocks, one May be set.

Further, when a common relative reference value is set in a plurality of blocks as described above, more accurate encoding can be performed by performing the following.

For example, if a relative reference value of 1 is set for 4 blocks, the shiftable bit amount S fij is 2 for 3 blocks and the shiftable bit amount S f U is 1 for one block. Suppose it was 5. In this case, only the block with the shiftable bit amount Sf ij of 5 stores the fact that it is further shifted by 3 bits. Specifically, a flag indicating that such special processing has been performed, its block number, and the number of additional shift bits are recorded. Note that, instead of the number of additional shift bits, the shiftable bit amount Sfij of only that block may be stored. Performing such special processing lowers the compression ratio. Therefore, for example, a threshold value may be determined for the decreasing compression ratio, and the determination may be made based on whether or not the data amount does not exceed the threshold value.

Note that the block group for setting the relative reference value of 1 is varied according to the spatial frequency of the image to be encoded (the number of blocks is small when the spatial frequency is high, and the number of blocks is low when the spatial frequency is low). May be increased).

Further, the first and second embodiments can be combined. For example, each block configuration data e 00 (0) to e 00 (15) shown in FIG. 14D is quantized by a predetermined number (for example, 2 bits), and the obtained shift block configuration data is made relative, A reversible left shift operation may be performed to perform vector quantization. Regarding the reversible left shift operation, as described above, the shift operation is performed by the shiftable bit amount Sfij indicating how many 0 bits are consecutive from the most significant bit MSB.

5. About creating a codebook

In the above embodiment, it is desirable that the codebook store block data corresponding to the image data to be compressed. In the following, the process of determining codebook candidate data (learning process) will be described.

In the present embodiment, a self-organizing feature map developed by Kohonen (T. ohonenn), which is one of the neural network algorithms, is used. Using the algorithm, we created a codebook (see Fig. 15) as follows.

First, create a total of 256 blocks from 00 to if (0 to 255). Such creation may be determined, for example, by generating a random number for each block candidate data constituting the block data. Note that the range of random numbers may be determined according to the bit length of the block configuration data stored in the codebook. In this case, in the coat book K (see Fig. 15), since each block configuration data is 4 bits long, random numbers may be generated in the range of 0 to 15.

It is assumed that the index number given to each block is specified by ij coordinates in a two-dimensional space as shown in FIG. In this case, since 256 blocks of data exist in the codebook P, the index number In can be specified by the two-dimensional coordinates of i = 0 to 16,〗 == 0 to 16. For example, it is expressed as an index number In02.

Also, the CPU 23 reads out the still image data from the hard disk 26, stores it in the memory 27, divides it into blocks of 4 * 4 pixels, and creates block data to be tested. deep. In this case, the block data T eOO to be tested is expressed as follows.

T e00 = (te00 (0), te00 (l), ..., t e00 (15))

A 4-bit shift process is performed to the right in order to make the block data T e 00 coincide with the bit length of the block configuration data used in the codebook P (step ST391 of FIG. 22). As a result, the following small shift block data Te00 'is obtained.

TeOO '= (t e00 (0) «\ t e00 (l)« \ · ·, t e00 (15) « ⁴ )

Next, the CPU 23 encodes the shift block data TeOO 'with the codebook K to be learned and obtains the encoded data <step ST393). As in the first embodiment, such encoding is determined by obtaining the total deviation TO for each block data of index numbers In00 to Inff and selecting the smallest one.

Next, the obtained encoded data is decoded using the same codebook K. / 1

(Step ST 395). This is code decoding block data U00. Code decoding block data U00 is represented by the following equation.

U00 = (u00 (0), uOO (l), · · · ~, u00 (15))

Next, the CPU 23 determines a block data to be learned (step ST397). In the present embodiment, as described below, block data having an index number whose distance from the block data specified by the coded data is equal to or smaller than a predetermined value is determined as block data to be learned. I did it.

If the index number In 43 shown in FIG. 21 is the block data identified by the encoded data, a square Sq with one side centered on the index number In 43, that is, the index number In 43 Are determined within a predetermined distance from. Then, block data located in the square SQ is set as a learning target. That is, 10 block data with index numbers In32, In33, In34, In42, In43> In44, In42, In52, In53, In54 are the block data to be learned. Becomes

Although the value of α is an arbitrary value, it is large in the early stage of learning and is reduced as the learning progresses. In other words, as learning progresses, the number of blocks to be learned decreases.

Next, CPU 23 performs a learning process on the block data to be learned (step ST 399). In the present embodiment, the learning process is performed as follows.

First, a learning process is performed on the first block data to be learned. In this case, the first block data to be learned is the block data identified by index number I η32. Assuming that this is block data # 32, block data # 32 is expressed as follows.

K32 = (k32 (0), k32 (l), ..., k32 (15))

Each data constituting the block data K32 is 4-bit data. On the other hand, the shift block T eOO 'which converted the test target block data T eOO

TeOO * = (t e00 (0) «\ t ε00 (1)« \ · ·, t e00 (15) « ⁴ ) 71

It is represented by

The difference between the two blocks is multiplied by a factor of 3 and subtracted from the block to be learned. That is, the block data L32 after learning is represented as follows.

L32 = K32- / 3 ■ (K32—TeOO ')

(Where 0 is 1)

That is,

L32 = (1-/ 3) -K32- j3 -TeOO '

Specifically, the block configuration data L32 (0) ′ to L32 (15) ′ forming the block data L32 are obtained as follows.

L32 (0) '= (1 -J3)-32 (0)-J3-t e 00 (0)

L32 (l) '= (1 -β) · Κ32 (1)-/ 3- t eO0 (l) «"

L32 (15) '= (1-j3) -K32 (15) -0

Such processing is performed on the block data specified by other index numbers In33, In34, In42, Iri43, In44, In42, In52, In53, and In54.

In this case, among the block data located in the square SQ, the block data T eOO to be tested and the block data having a larger error have a larger subtraction value in the learning process, and the block data having a smaller error have a larger error. The smaller the subtraction value in the learning process, the smaller the value. As a result, the block data located in the square SQ converges to certain block data.

The coefficient 3 is an arbitrary value, like the coefficient α, but is large at the beginning of the learning and is reduced as the learning progresses. In other words, as the learning progresses, the error correction value decreases.

Thus, the learning process using the block configuration data TeOO to be tested ends. The learning process, that is, the learning process using another block configuration data of the test object Te OK Te02 ♦ An appropriate codebook can be created from the generated codebook.

In the above explanation, the number of block data created initially is set to 256 for simplicity of explanation, but in the initial state, it is increased to 512, 1026, etc. Then, the converged block data may be combined into one.

In addition, although the square SQ having one side α is defined for determining the block data to be explained in the above description, a circle or another polygon may be defined. In the above description, the index numbers assigned to the respective blocks are determined by ij coordinates in a two-dimensional space as shown in FIG. 21. However, the present invention is not limited to this. It may be determined as a position in a dimension and further in a multidimensional space. If a three-dimensional space is considered, a region within a predetermined distance from the index number In 43 may be a solid.

In the first embodiment, the data is divided into groups 0 to 7 based on the difference data. Therefore, test target block configuration data having the difference data matching each group may be provided.

In the above embodiment, the learning process in step ST399 of FIG. 22 is performed so that all the block data located in the square SQ are converged. However, it can also be processed as follows.

The area in the square SQ is divided into two areas, an area near the center S Q i and an area far from the center S q o. For the block data belonging to the region S qi near the center, a process of subtracting the difference between the two block data and the coefficient) 3 from the learning target block data is performed. That is, the block data xy 'after learning is represented as follows.

K xy '= Kxy-β · (Kxy-T e 00')

(Where 0 <<1)

On the other hand, for the block data belonging to the region Sqo far from the center, a process of multiplying the difference between the two block data by a coefficient β and adding the result to the block data to be learned is performed. That is, the block data Kxy 'after learning is expressed as follows.

— 3 2 —

Replacement form (Rule 26) . Kxy '= Kxy + / 3-(Kxy-T eOO')

(Where 0 <j3 <l)

By performing such processing, it is possible to converge those near the center and to move away from those far from the center. Thereby, convergence can be achieved faster.

In addition, those that are close to the center are converged, and those that are far from the center are not limited to subtraction processing and addition processing. The value of coefficient) 3 may be changed for the region of.

In the present embodiment, the Kohonen neural network is used as the learning process for the codebook candidate data. May be adopted. Further, a plurality of learning methods may be combined.

[D: About Hadamard transform]

In each of the above embodiments, the block configuration data may be subjected to Hadamard conversion. For example, in the first embodiment, the shift block configuration data A00 ′ shown in FIG. 14A may be converted by a Hadamard matrix as shown in FIG. 23 between step ST19 and step ST21 in FIG. In this case, in the decoding process, the inverse Hadamard transform may be performed between step ST37 and step ST38 in FIG.

As described above, by performing the Hadamard transform before the vector quantization, a more accurate transform can be performed. This is because the DC component in each block can be concentrated by the Hadamard transform. Note that an orthogonal transform other than the Hadamard transform, such as DCT, may be used.

Also, by performing Hadamard transformation after the relativization processing, the amount of calculation hardly increases, and the resolution can be increased. This is because the low-frequency component relative treatment, will almost gone, _c thus that is because it is possible to increase the resolution of the high-frequency components, Hadamard transform, removes the low-frequency components of a block configuration data It is desirable to perform from.

As described above, in the first embodiment, the Hadamard transform is performed after the relativization process, and then the vector quantization is performed.

In the second embodiment, the obtained block configuration relative data may be subjected to the Hadamard transform between the step ST257 and the step ST259 in FIG. In this case, in the decoding process, inverse Hadamard transform may be performed between step ST2777 and step ST278 of FIG.

It should be noted that the Hadamard transform is not performed before the reversible left shift operation (step ST2661) but is performed after the reversible leftshift operation (between step ST2661 and step ST2663). Is also good. In this case, in the decoding process, the inverse Hadamard transform may be performed between step ST177 and step ST177 in FIG. 20. Note that in this specification, the Hadamard transform is a fast Hadamard transform. Including conversion. 6. Other embodiments

In the present embodiment, the number of blocks having a high evaluation value is determined as the number of higher-ranked blocks, and coding is performed. As a result, a fixed number of blocks of data can always be transmitted as a block to be encoded, so that blocks that can be transmitted per unit time regardless of the size, speed, number of moving objects, etc. Can be kept constant. Therefore, when the transmission speed is limited, only the important data can be transmitted.

In this embodiment, the evaluation values are sequentially extracted in descending order. Therefore, when the evaluation value as a whole is large and when the evaluation value is small as a whole tendency, even if the blocks have the same evaluation value, some blocks are not transmitted because of relative evaluation. On the other hand, if the whole moves automatically as a whole, it is possible to transmit only the parts that have moved significantly. On the other hand, if the whole does not move much, the fine movement will be transmitted.

Instead of determining a block to be coded by a relative evaluation value, a block having a difference evaluation value which is more than an absolute value may be transmitted.

— 3 4 —

Replacement paper (Rule 26) Also, instead of limiting the number of blocks to be transmitted, it may be determined that a block having a value of what percentage of the largest differential evaluation block in the frame is transmitted.

In the present embodiment, the 50 blocks are sequentially extracted from the block having the larger difference evaluation value, and the extraction operation is stopped at the 50 blocks. If there are two or more blocks having the 50th difference evaluation value, one of the blocks having the 50th difference evaluation value is extracted while observing the restriction of 50 blocks. What should I do? In this case, for example, one having a smaller block ID may be preferentially extracted. If there are too many blocks having the 50th difference evaluation value, extraction of the block having the 50th difference evaluation value is stopped, and coding is performed on 49 blocks. What should I do? That is, when blocks having the same difference evaluation are over the threshold value, the balance between blocks to be extracted and blocks not to be extracted may be considered.

Further, the threshold value may not be a fixed value but may be varied according to the distribution of the difference evaluation value of the block data. For example, for a certain frame, the number of blocks to be transmitted is changed depending on whether there is little movement as a whole compared to the previous frame or when there is considerable movement as a whole. As a result, useless data transfer can be prevented, and transmission more suitable for the characteristics of the image data to be encoded becomes possible.

As described above, in the case where the distribution is varied according to the distribution of the difference evaluation value of each block, a histogram may be created to easily grasp the distribution state. In this case, a histogram may be created based on how many blocks having a difference evaluation value belonging to a certain range. For example, if each pixel has a filter level of 0 to 255 and 1 block is composed of 16 pixels, the range of the difference evaluation value is 0 to 4 0 9 6 (2 5 5 * 1 6) This can be divided into 32 ranges, and the difference evaluation values 0 to 1 27 can be classified into the first range, the difference evaluation values 1 28 to 255 can be classified into the second range, and so on. . By analyzing the distribution of the histogram, the distribution of the difference evaluation value can be obtained. When such a histogram is created, the difference evaluation value of each block is classified according to the range, so that there are many cases where there are a plurality of blocks having the 50th difference evaluation value as the threshold value. Become. However, in this case as well, processing may be performed in the same manner as when there are a plurality of blocks having the same difference evaluation across the threshold value.

Alternatively, the transmission state may be detected, and the threshold value may be changed based on the difference evaluation and the detection result. For example, if there are many errors during transmission, the threshold may be increased to reduce the number of blocks to be transmitted. This makes it possible to flexibly cope with a case where the transmission state fluctuates.

Also, if multiple transmission lines are available, use multiple transmission lines if they are moving significantly as a whole, and use only one transmission line if they are not moving very much. By changing, the transmission condition may be changed.

Also, the encoding method can be changed as a change in transmission conditions. For example, a codebook for performing rough coding may be prepared, and coding may be performed at a high compression ratio with reduced accuracy for a portion having a small difference evaluation value.

In this embodiment, data compression is performed by performing vector quantization using the codebook shown in FIG. 15 and the like, but other DCT (discrete cosine transform) or the like is used. Compression may be performed. That is, any compression method that can perform data compression for each predetermined block can be used.

Further, in this embodiment, one block is composed of 4 × 4 16 pixels. However, the present invention is not limited to this, and an arbitrary number of pixels such as 8 × 8 can be used.

In the above embodiment, the transmission method is changed by extracting transmission target block data from a plurality of block data to be encoded based on the difference evaluation of each block data. However, besides this, the transmission method may be changed by changing the transmission conditions for the block data to be transmitted based on the difference evaluation between the blocks.

Also, in the above embodiment, the block having the larger difference evaluation is used as the block to be encoded.

— 3 6 —

Replacement form (Rule 26) To be decided as However, the order of transmission may be larger. As a result, important data can always be transmitted. So, for example, if the bandwidth changes during transmission

Also, if the receiving side is performing multitask processing, less important blocks (small differential evaluation) will not cause much problem even if they are ignored as a result. Conversely, if the bandwidth becomes wide, a block with a small difference evaluation may be transmitted.

The threshold value may be set so that the operator can freely select it. For example, it may be possible to switch between the case where the precise mode is specified, the case where the speed is prioritized, and the like. That is, in the case of the precision mode, the threshold value may be lowered, and in the case of speed priority, the threshold value may be raised.

Further, in the above embodiment, all the transmission targets are encoded and transmitted. That is, encoding is not performed for those that are not to be transmitted. However, after all coding, only those having a large evaluation value may be transmitted. Alternatively, a difference evaluation may be obtained after the encoding process to determine whether or not to perform transmission.

In the above embodiment, the description has been made assuming that data is transmitted. However, when the data is stored in a storage medium, the present invention can be applied as a device for compressing and storing data. In this case, the directly encoded block data may be stored in the storage medium instead of being transmitted from the transmission control unit.

The video data transmission device and the reception device according to the present invention can transmit and receive video data without significantly deteriorating the accuracy of images and without increasing the data transfer amount. It can be applied to systems and the like.

In the above embodiments, the digital image data has been described as being obtained. However, such digital image data can be obtained, for example, by sampling analog still image data.

When the digital image data is a single color image, the same processing as described above may be performed for each of the YUV signals (the luminance signal Y and the hue signal UV).

1 3 7 ―

Replacement form (Rule 26) Instead of the YUV signal, an RGB (red, green, blue) signal or a YIQ (luminance signal Y and hue signal IQ) may be used.

It is also known that human vision reacts sharply and sensitively to details of a given shape in light and dark, but responds slowly to details of color. Therefore, in the present embodiment, the resolution of the hue signals U and V is set to 1/2 of the luminance signal Y. As a result, the degree of data compression can be further increased.

In each of the above embodiments, a case where compressed data is transmitted has been described. However, the present invention can be applied to a case where data is stored as it is.

In the embodiment, the value of the difference data Aijg increases if one pixel has the maximum value or the minimum value. That is, when there is an error such as dust, there is a possibility that compression is performed using a codebook different from the original one. Therefore, in order to solve such a problem, for example, instead of the difference data A ijg = maximum value−minimum value, the difference data A ijg is calculated from (an average value of several pixels having large values) Alternatively, it may be obtained by subtracting the average value of several small pixels). In the first embodiment, in the first embodiment, 8 groups are set according to the value of the difference ijg. However, the number of groups may be increased or decreased.

In the above embodiment, the difference evaluation is obtained by summing the absolute values of the differences between the pixels in the frame buffer corresponding to the comparison buffer. However, the present invention is not limited to this, and the total number of pixels having a difference may be calculated as the difference evaluation in the block.

In each of the above embodiments, each of the above functions is realized by software using CPU 23. However, a part or all of them may be realized by a piece of hardware such as a logic circuit. In particular, in the present embodiment, encoding and decoding are performed using a codebook. Therefore, by realizing the encoding and decoding processes using one chip, high-speed and high-precision processing can be achieved simply by embedding the chip in devices that require decompression, such as videophones and digital video devices. High compression is possible.

In the above, the present invention has been described as a preferred embodiment.

Replacement form (Rule 26) It has been used for purposes of explanation, not for clarity, and may be modified within the scope of the appended claims without departing from the scope and spirit of the invention.

[F: Definition of terms]

Hereinafter, the meaning of terms used in the present specification and the relationship with the embodiments will be described.

"Transmission method": Includes all methods of transmitting a plurality of block data that make up one screen. For example, the block data that is to be transmitted and the other block data out of a plurality of block data that make up one screen Of course, the transmission conditions such as the transmission order, encoding method, transmission speed, and transmission path of the block data to be transmitted are also included.

"Encoding reference image data overnight storage means": In the embodiment, the comparison buffer 27a shown in FIG. 5 corresponds to this.

"Encoding transmission means": In the embodiment, the processing of CPU23 in FIG. 6 step ST109, FIG. 7 step ST123, step ST127 is applicable.

"Difference evaluation calculation means": means for calculating a difference evaluation in each block, and corresponds to the processing of step ST113 of FIG. 6 of CPU23 in the embodiment.

<< Encoding reference image data rewriting means >>: In the embodiment, this corresponds to the processing of step ST125 of FIG.

“Block configuration data”: Refers to data configuring block data. In the embodiment, 16 block configuration data is used to represent one block data.

“Relative block configuration data”: Refers to block configuration data represented by relative values. "Relative block data storage means": In this embodiment, a hard disk 26 useful for the code book K in FIG.

“Conversion reference value”: data for converting each block configuration data of the block data to be encoded into a relative block data represented by a relative value. The minimum value of the block configuration data that composes the block data of was adopted.

ΓCandidate data ”: Consists of block data and the corresponding code You. In the first embodiment, this corresponds to the block configuration data of the codebook K shown in FIG. 15 and the corresponding index numbers.

"First conversion means": Corresponds to the processing in step ST19 of FIG. 12 of the CPU23. "Relative block data encoding means": Corresponds to the processing of ST21 in Fig. 12 of CPU23.

"Block characteristic data overnight calculation means": Corresponds to the processing of step ST15 and step ST17 in Fig. 12 of CPU23.

“Correspondence table by group”: Codebook K corresponds to grouping according to block characteristic data.

"Group name j: This is a name for specifying which group in the codebook belongs to which relative block data is used for encoding. The name here is a concept that includes IDs including numbers and the like. It is.

“Group number”: A name for specifying which relative block data belonging to a certain group in codebook K is used for encoding. Here, the intra-group number is a concept including ID including a name and the like.

“Difference data”: Difference data between the maximum value and the minimum value of each block configuration data constituting the block data to be encoded. In the embodiment, the maximum value ma X Ai j 'one minimum It is determined by the value minAij '.

{2nd conversion means}: Corresponds to the processing of step ST37 in FIG. 18 of the CPU 23 in the embodiment.

“Relative block data decoding means”: Corresponds to the processing in step ST35 in FIG. 18 of the CPU 23 in the embodiment.

"Orthogonal transform J: In the embodiment, the Hadamard transform corresponds, but other orthogonal transforms include, for example, DCT.

"Inverse orthogonal transformation": Performing the inverse processing of the orthogonal transformation. In the embodiment, the inverse Hadamard transformation corresponds. :

ΓReversible left shift operation ”: This refers to performing a left shift operation on the largest value block configuration data of each block configuration data until the most significant bit becomes 1 for each predetermined block. In the embodiment, for example, Figure 19 Step ST 26 1 processing Applicable.

“Reversible right shift operation”: Shifting to the opposite side from the reversible left shift. In the embodiment, for example, the process of step ST277 in FIG.

In the moving surface data encoding apparatus or method according to the present invention, a difference evaluation between block data constituting the image data to be encoded and corresponding block data of the encoded reference image data is calculated. Then, based on the difference evaluation of each block data, the transmission method of the given encoding target image data is changed and transmitted. Therefore, transmission according to the characteristics of moving image data can be performed.

In the moving picture data encoding apparatus according to the present invention, the encoding transmission means may transmit the block data to be transmitted from a plurality of encoding data to be encoded based on the difference evaluation of each block data. Extract. Therefore, some block data can be excluded from transmission targets based on the difference evaluation. As a result, transmission according to the characteristics of moving images can be performed.

In the moving picture data encoding device according to the present invention, the encoding reference image data rewriting means rewrites the block data of the encoding reference screen data using the transmission target block data. Accordingly, encoded reference image data in which a portion corresponding to the transmission target block data is rewritten is obtained.

In the moving picture data encoding apparatus according to the present invention, the encoding transmission means changes transmission conditions for the block data to be transmitted based on the difference evaluation of each block data. This enables transmission according to the characteristics of moving image data. In the moving picture data encoding device according to the present invention, the encoding transmission unit changes the transmission method according to the evaluation value distribution of each difference evaluation of the block data of the image data for one screen. Therefore, transmission according to the characteristics of each screen becomes possible.

In the moving picture data encoding device according to the present invention, the encoded transmission unit detects a transmission state, and changes the transmission method based on the difference evaluation and the detection result. Therefore, the transmission method can be changed according to the transmission state. In the moving picture data encoding device according to the present invention, the encoding transmission unit compares the transmission target block data by comparing with a preset difference threshold value.

— 4 I — 0 T / JP 7 1

Extract the evening. Therefore, a block having a difference evaluation exceeding the difference threshold can be determined as a transmission target. As a result, only block data having large fluctuations can be transmitted.

In the moving image data encoding apparatus according to the present invention, the encoding transmission unit may include a block having a predetermined ratio of blocks irrespective of the evaluation value distribution of each difference evaluation of the block data of the image data for one screen. The transmission target block data is extracted from the data. Therefore, a certain percentage of block data can be transmitted. Thereby, the transmission rate and the like can be kept constant.

In the moving picture data encoding device according to the present invention, the encoding transmission unit: 1) determines a conversion reference value for each of a predetermined number of blocks for the block data to be encoded, and configures the predetermined number of blocks. First conversion means for converting each block configuration data to be converted into relative block data represented by a relative value with respect to the conversion reference value; and 2) candidate data in which the converted relative block data is stored in advance. And a relative block data encoding means for outputting encoded data for each block and outputting the conversion reference value for each predetermined number of blocks. As described above, by converting the data into the relative block data using the relative reference value for each of the predetermined number of blocks and encoding the data, high-precision compression is possible even if the stored candidate data is lost. Becomes Also, by reducing the number of candidate data to be stored, high-speed encoding becomes possible.

In the moving picture decoding apparatus according to the present invention, each of the block configuration data is represented by a binary number, and the relative block decoding processing means further includes: Before coding, for each predetermined block, a shiftable bit amount indicating how many 0s are connected from the most significant bit in the block configuration data having the maximum value among the block configuration data is obtained, A reversible left shift operation is performed on each block configuration data on the basis of the shiftable bit amount for each predetermined block, the encoding is performed, and the shiftable pit amount is output for each predetermined block. Therefore, useful candidate data can be reduced without reducing accuracy.

In the moving picture data encoding device according to the present invention, the predetermined number of blocks is one block. Therefore, a relative value can be determined for each block. In the moving picture data encoding device according to the present invention, the first conversion means performs an orthogonal transform when converting the block configuration data into the relative block configuration data. Therefore, more accurate conversion can be performed.

In the moving picture data encoding device according to the present invention, the first transforming unit performs the orthogonal transform after performing a relative block de-conversion. Therefore, more accurate conversion can be performed.

In the moving picture data encoding device according to the present invention, the first conversion unit converts the block configuration data constituting the block data to be encoded into the relative block data after converting the block data into a relative block data. Perform a shift operation. As described above, the reversibility is shifted leftward after relativization, so that the resolution can be increased.

In the moving picture decoding apparatus according to the present invention, the first conversion unit performs the reversible left shift operation after the orthogonal transformation. In this way, after relativization, reversibility is left-shifted and orthogonally transformed, so that the resolution can be increased.

In the moving picture data encoding device according to the present invention, a plurality of relative block data composed of a plurality of block configuration data represented by relative values and a code corresponding to the relative block data are stored as candidate data. A conversion reference value in the block data is determined based on the block data to be converted, and each block configuration data included in the block data is converted into relative block data represented by a relative value to the conversion reference value. . Then, the relative block data is encoded using the candidate data stored in the relative block data storage means, and the relative code data and the conversion reference value are output.

As described above, by converting to the relative block data using the relative reference value and encoding the data, highly accurate compression can be performed even if the number of candidate data to be stored is reduced. Also, by reducing the number of pathological candidate data, high-speed encoding becomes possible.

In the moving picture data decoding apparatus and the moving picture data decoding method according to the present invention, when the coded block data obtained by coding the block data and its block ID are given, the coded block data is decoded. A frame next to the reference image data is generated based on the reference image data and the decoded block data. The image data of the next frame is generated, and the encoded reference image data is rewritten to the next frame image data. Therefore, it is possible to receive the encoded data transmitted from the moving image data encoding apparatus according to the present invention and decode it. In the moving picture decoding apparatus according to the present invention, the decoding means stores a plurality of the relative block data and codes corresponding thereto as candidate data, and stores the relative code data. The decoding is performed using the stored candidate data, and the decoded relative block data is converted into the relative block data before encoding based on the conversion reference value in the block data. Therefore, when the relative code data and the conversion reference value are provided, they can be decoded.

In the moving image decoding apparatus or the decoding method according to the present invention, the difference evaluation between the block data constituting the image data to be encoded and the corresponding block data of the encoding reference image data is performed. Then, block data to be encoded is determined based on the difference evaluation of each block data. Therefore, encoding according to the characteristics of moving image data can be performed.

In the moving picture data decoding device or the decoding method thereof according to the present invention, when the coded block data obtained by coding the block data and the block ID thereof are given, the coded block data is decoded. Based on the reference image data and the decoded block data, image data of a frame next to the reference image data is generated. Then, the encoding reference image data is rewritten to the next frame image data. Therefore, it is possible to receive the encoded data transmitted from the moving image data encoding device according to the present invention and decode the encoded data.

4 4

Replacement paper (Kaikai IJ26)

Claims

Claims: Encoded reference image data storage means for storing one screen of image data composed of a plurality of pixels as encoded reference image data,

Given image data to be encoded is divided into a plurality of block data, and for each of the divided block data, a block data different from the corresponding block data of the encoding reference image data is encoded. Encoding transmission means for transmitting the encoded block data together with the block ID,

In the moving image data encoding device provided with

A differential evaluation operation unit for calculating a difference evaluation between block data constituting the image data to be encoded and corresponding block data of the encoding reference image data, wherein the encoding transmission unit comprises: Changing the transmission method of the given image data to be encoded based on the difference evaluation of

A moving image data encoding device characterized by the following.

2.

The moving image data encoding device according to claim 1,

The encoding transmission unit changes the transmission method by extracting a transmission target block data from a plurality of block data to be encoded based on the difference evaluation of each block data,

A moving image data encoding device characterized by the following.

3.

The moving picture data encoding device according to claim 2,

An encoding reference image data rewriting unit that rewrites the block reference image of the encoding reference screen data using the transmission target block data.

A moving image data encoding device characterized by the following.

Five -

Four .

The moving image data encoding device according to claim 1,

The encoding transmission means changes the transmission method by changing transmission conditions for the block data to be transmitted based on the difference evaluation of each block data;

A moving image decoding apparatus characterized in that:

Five .

The moving picture data encoding apparatus according to claim 1,

The encoding transmission unit changes the transmission method according to an evaluation value distribution of each difference evaluation of block data of the image data for one screen,

A moving image data encoding device characterized by the following.

6.

The moving image data encoding device according to claim 1,

The encoded transmission unit detects a transmission state, and changes the transmission method based on the difference evaluation and the detection result;

A moving picture data encoding device characterized by the following.

7.

The moving picture data encoding device according to claim 2,

The encoded transmission unit extracts the transmission target block data by comparing with a preset difference threshold value;

A moving picture data encoding device characterized by the following.

8.

The moving picture data encoding device according to claim 4,

The encoded transmission unit changes the transmission condition by comparing with a preset difference threshold value; Moving image data encoding apparatus.

9.

The moving picture data encoding device according to claim 2,

The encoding transmission means extracts the transmission target block data for a predetermined percentage of block data regardless of the evaluation value distribution of each difference evaluation of the block data of the image data for one screen. thing,

A moving image data encoding device characterized by the following.

Ten .

The moving picture data encoding device according to claim 4,

The coded transmission means changes the transmission condition for a preset ratio of block data regardless of the evaluation value distribution of each difference evaluation of the block data of the image data for one screen. ,

A moving image data encoding device characterized by the following.

In the moving image data encoding device according to any one of claims 1 to 10,

The coded transmission means includes the following means 1) to 2),

1) For the block data to be encoded, a conversion reference value is determined for each of a predetermined number of blocks, and each block configuration data constituting the predetermined number of blocks is determined as a relative value with respect to the conversion reference value. First conversion means for converting to relative block data represented by

2) The relative block data after the conversion is coded by candidate data stored in advance, and coded data is output for each block, and the conversion reference value is output for each predetermined number of blocks. Encoding means,

A moving image data encoding device characterized by the following.

1 2.

The moving image data encoding device according to claim 11, Each of the block configuration data is represented by a binary number,

The relative block data encoding means may further include, before performing the encoding for each block data, for each predetermined block, the most significant bit in the block configuration data having the maximum value among the block configuration data. , A shiftable bit amount indicating how many bits are consecutive from 0 is calculated, a reversible left shift operation is performed on each block configuration data based on the shiftable bit amount for each predetermined block, and the encoding is performed. Output the amount of shiftable bits for each predetermined block,

A moving image data encoding device characterized by the following.

13 .

The moving image data encoding device according to claim 12,

The predetermined number of blocks is one block,

A moving image data encoding device characterized by the following.

14 .

The moving image decoding apparatus according to any one of claims 11 to 13, wherein the first conversion unit performs orthogonal transformation when converting the block configuration data into the relative block configuration data.

A moving image data encoding device characterized by the following.

1 5.

The moving image data encoding device according to claim 14,

The first converting means performs the orthogonal transform after converting the data into relative block data;

A moving image data encoding device characterized by the following.

1 6.

The moving image data encoding device according to claim 15,

The first converting means converts the block to be encoded into

— 8 —

Replacement form (Rule 26) Subjecting each block configuration data constituting the lock data to the reversible left bit shift operation,

A moving image data encoding device characterized by the following.

1 7.

The moving image data encoding device according to claim 15,

The moving image data encoding device, characterized in that the first transforming unit performs the reversible left shift operation after the orthogonal transform.

1 8.

A) encoding reference image data storage means for storing one screen of image data composed of a plurality of pixels as encoding reference image data;

B) Given encoded block data obtained by encoding the block data and its block ID, decoding means for decoding the encoded block data;

D) encoding reference image data rewriting means for rewriting the encoding reference image data of the encoding reference image data useful means to the next frame image data;

A moving image data decoding device comprising:

1 9.

The moving picture data decoding device according to claim 18,

The decoding means is provided with relative code data obtained by encoding relative block data converted using a conversion reference value of block data composed of a plurality of block configuration data, and a conversion reference value in the block data. And decoding it by the following means 1) to 3), G relative block data storage means for storing a plurality of relative block data and a code corresponding thereto as candidate data,

2) relative block data decoding means for decoding the relative code data using candidate data stored in the relative block data recording means,

3) second conversion means for converting the decoded relative block data into relative block data before encoding by using a conversion reference value in the block data;

2 0.

Image data of one screen composed of a plurality of pixels is recorded as encoding reference image data, and a given image data to be encoded is divided into a plurality of block data, For each of the divided block data, block data different from the corresponding block data of the coding reference image data is coded as block data to be coded, and the coded block data of the coded block is transmitted together with its block ID. Moving image data encoding method,

The difference evaluation between the block data constituting the image data to be encoded and the corresponding block data of the encoding reference image data is calculated, and based on the difference evaluation of each block data, the given encoding target is calculated. Changing the transmission method of the image data of

A moving image data encoding method characterized by the following.

twenty one .

A moving image data decoding method for receiving encoded data and decoding the encoded data, wherein an image data for one screen composed of a plurality of pixels is stored as a reference image data,

Given an encoded block data that encodes block data and its block ID, the encoded block data is decoded,

Based on the reference image data and the decoded block data, generate image data of a frame next to the reference image data, Rewriting the coded reference image data with the next frame image data.

twenty two .

Encoding reference image data storage means for storing image data for one screen composed of a plurality of pixels as encoding reference image data;

The given encoding target image data is divided into a plurality of block data, and for each divided block data, a block data different from the corresponding block data of the encoding reference image data is encoded. Encoding means for determining as block data to be encoded, and encoding the block data corresponding to the block ID;

In the moving image data encoding device provided with

The image processing apparatus further includes a difference evaluation operation unit configured to calculate a difference evaluation between the block data configuring the image data to be encoded and the corresponding block data of the encoding reference image data. Determining the block to be encoded based on the difference evaluation;

A moving image data encoding device characterized by the following.

twenty three .

A moving picture data decoding apparatus for decoding the coded data encoded by the moving picture data coding apparatus,

A) reference image data storage means for storing one screen of image data composed of a plurality of pixels as reference image data;

B) Decoding means for decoding the block data after the block data after the block data is encoded and the block ID thereof are given;

C) based on the reference image data and the decoded block data, a next frame image data generation means for generating image data of a frame next to the reference image data,

— 5 I — Replacement Form (Rule 26) A moving picture data overnight decoding device comprising:

twenty four .

One screen of image data composed of a plurality of pixels is stored as encoding reference image data, and given image data to be encoded is divided into a plurality of block data, and each of the divided block data is encoded. In the moving image data encoding method, the block data different from the block data corresponding to the encoding reference image data is encoded as the encoding target block data in association with the block ID. Calculates the difference evaluation between the block data constituting the image data and the corresponding block data of the coding reference image data, and based on the difference evaluation of each block data, the block to be coded. Determining data,

A moving picture data encoding method characterized by the following.

twenty five .

A moving image data decoding method for decoding encoded data,

One screen of image data composed of a plurality of pixels is stored as encoding reference image data,

Given encoded block data obtained by encoding block data and its block ID, the encoded block data is decoded,

Rewriting the coded reference image data with the next frame image data.

2 6.

A computer-readable storage medium storing a computer-executable program, wherein the program implements the device or method according to any one of claims 1 to 25.

A storage medium characterized by the above-mentioned.

Replacement form (Rule 26)