JP2001169291A - Device and method for compressing data, and storage medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the compressibility of a moving picture while the quality of a reproduced picture is maintained. SOLUTION: This device is provided with a motion vector detecting part 12 for detecting a motion vector by using two frames constituting the moving picture, a motion vector picture reproducing part 13 for generating a predictive picture in a present frame by using the detected motion vector, a difference picture arithmetic part 14 for generating a difference picture by calculating a difference between the generated predictive picture and a picture in the present frame inputted from a picture input part 11, and a difference picture vector quantization part 15 for executing vector quantization to the difference picture. Then, the device executes vector quantization to the difference between the actual picture in the present frame and the predictive picture predicted from the motion vector so as to execute the data compression of the moving picture by high compressibility while the deterioration of the picture is suppressed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data compression device, a compression method, and a storage medium storing a program for executing these processes, and more particularly, to a compression device using vector quantization as one of data compression techniques. It is suitable for use as a compression method and a storage medium.

[0002]

2. Description of the Related Art Conventionally, various data compression techniques have been proposed. Among them, a method called “vector quantization” is often used as one of data compression algorithms that can very easily perform decompression processing of compressed data. This algorithm has been known for a long time in the field of signal processing, especially for data compression of image signals and audio signals,
Or it has been applied to pattern recognition.

In this vector quantization, several pixel patterns of a certain size (for example, a block of 4 × 4 pixels) are prepared, and a unique number or the like is given to each of them (this set is called a “codebook”). ). Then, for example, blocks of the same size (for example, 4 × 4 pixels) are sequentially extracted from the image data of the two-dimensional array, a pattern most similar to the block is found in the codebook, and the pattern number is assigned to the block. Perform data compression. In vector quantization, a data sequence in one block corresponds to one code vector.

On the receiving side or decompressing side of the coded compressed data, the original image can be reproduced simply by extracting the pattern corresponding to the number for each block from the code book. Therefore, on the extension side,
If a codebook is received or stored in advance, no special calculation is required, and the original image can be reproduced with very simple hardware.

[0005] Data compression by vector quantization is applied not only to still images but also to moving images. A moving image is a set of a plurality of still images (frames) that change with the passage of time, and data of each moving image can be compressed by compressing the data of each still image forming the moving image by vector quantization.

[0006]

However, even if each frame constituting a moving image is compressed by vector quantization, the compression rate is equivalent to the vector quantization of a still image. There was a limit. Also,
Even if data of several frames constituting a moving image is compressed together, the compression ratio is limited, and it is difficult to improve the compression ratio while maintaining the quality of a reproduced image.

The present invention has been made to solve such a problem, and an object of the present invention is to improve the compression ratio of a moving image while maintaining the quality of a reproduced image.

[0008]

A data compression apparatus according to the present invention comprises a motion vector detecting means for detecting a motion vector using a plurality of frames constituting a moving picture, and a motion vector detected by the motion vector detecting means. Prediction means for generating a predicted image of the current frame by using the difference image generation means for calculating a difference between the predicted image generated by the prediction means and the image of the input frame to generate a difference image; A difference image vector quantization unit that performs vector quantization on the difference image generated by the image generation unit.

The data compression apparatus according to the present invention comprises a motion vector detecting means for detecting a motion vector using a plurality of frames constituting a moving picture, and a vector quantization for each of the plurality of frames constituting the moving picture. Vector quantization means for outputting a code string obtained for each macroblock in a frame, and a difference between the sum of the luminances of the two codevectors by comparing the codevectors of the positionally corresponding macroblocks between the two frames A luminance sum calculating means for calculating an absolute value and judging that code vectors having a difference absolute value of the luminance sum greater than a first threshold value are dissimilar to each other; Compute the code vectors of the macroblocks to calculate the spatial distance between the two code vectors, wherein the spatial distance is equal to a second threshold value. And spatial distance calculating means for determining dissimilarity between the large code vector between Ri, the code string of each frame output from the vector quantization means,
Output control means for outputting only a code string corresponding to both the code vector determined to be dissimilar by the luminance sum value calculating means and the code vector determined to be dissimilar by the spatial distance calculating means, Each frame constituting a moving image includes a predicted frame and a key frame arranged with at least one of the predicted frames interposed therebetween. When the current frame is the key frame, the current frame is temporally before or after the key frame. Using the other key frames, the vector quantization means, the luminance sum value calculation means, the spatial distance calculation means and the output control means perform the processing, and when the current frame is the predicted frame, the motion vector Only processing by the detecting means is performed, and only the obtained motion vector is output as compressed data.

In the data compression method according to the present invention, a first method for detecting a motion vector using a plurality of frames constituting a moving image is provided.
And a second step of generating a predicted image of the current frame using the motion vector detected in the first step; and a predicted image generated by the second step and an image of the input frame. And a fourth step of vector-quantizing the difference image generated in the third step by calculating a difference between the difference image and the difference image.

According to the data compression method of the present invention, a first method for detecting a motion vector using a plurality of frames constituting a moving image is provided.
And a second step of performing vector quantization on each of a plurality of frames constituting the moving image and outputting a code string obtained for each macroblock in the frame. The code vectors of the corresponding macroblocks are compared with each other, and the difference absolute value of the luminance sum of both code vectors is calculated, and the code vectors whose difference absolute value of the luminance sum is larger than the first threshold value are dissimilar. And calculating a spatial distance between both code vectors by comparing the code vectors of the positionally corresponding macroblocks between the two frames, and determining that the spatial distance is equal to a second threshold. A fourth step of judging code vectors larger than the value to be dissimilar, and a code string of each frame output in the second step A fifth step of outputting only a code string corresponding to both the code vector determined to be dissimilar in the third step and the code vector determined to be dissimilar in the fourth step. Each frame constituting the moving image includes a predicted frame and a key frame arranged with at least one of the predicted frames interposed therebetween. When the current frame is the key frame, the current frame is temporally preceding the key frame. Or the second step, the third step
And the processing of the fourth and fifth steps. When the current frame is the predicted frame, only the processing of the first step is performed, and only the obtained motion vector is compressed data. Output as

The storage medium of the present invention is a computer-readable storage medium storing a program for causing a computer to function as each unit of the data compression device.

The storage medium of the present invention is a computer-readable storage medium storing a program for causing a computer to execute the procedure of the above data compression method.

[0014]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Some embodiments of the present invention will be described below in detail with reference to the drawings. In each of the embodiments described below, a moving image is taken as an example of data to be compressed, and the compression ratio of image data and the reproduction quality are further improved.

(First Embodiment) First, a first embodiment of the present invention will be described with reference to FIGS. First, the principle of image compression and decompression by vector quantization will be described with reference to FIG. 3 taking a still image as an example. As shown in FIG. 3, an original image 1 input as data to be compressed is configured by a large number of elements called pixels. Each pixel has information such as a luminance signal (Y signal) converted from an RGB signal and a color signal (U, V signal).

An input image block (macro block) 2 is obtained by extracting a block composed of a plurality of pixels from the input image 1. In the example of FIG.
Although 4 × 4 pixels are selected as the size of, this size may be anything. Since the input image block 2 has a plurality of pixels as described above, the luminance signal values and the chrominance signal values of each pixel can be collected and used as vector data. This is the input vector data.

Due to human visual characteristics, some input image blocks in the input image 1 may look almost the same in appearance. Such a plurality of input image blocks that look the same can be represented by a smaller number of image blocks. The image block codebook 3 has a plurality of image blocks (code vector data) representing a large number of input image blocks on the input image 1.
Code vector data is stored in the image block code book 3
The luminance and chromaticity of each pixel of the image blocks in the image data are vector data.

In the vector quantization, the entire input image 1 is divided into image blocks, and each image block 2 is used as input block data, and code vector data similar to the input vector data is searched from the code book 3. Then, the image can be compressed by transferring only the corresponding code vector data number. In order to reproduce the compressed image and obtain the reproduced image 4,
The code vector data corresponding to the transferred number may be read from the code book 3 and applied to the image.

FIG. 4 shows a technique for compressing moving image data by this vector quantization. A moving image is configured by continuously switching a plurality of (about 60 frames) images per second. As shown in FIG. 4, first, the respective frames 5 to 7 of the original image are subjected to vector quantization (VQ) processing for each macroblock of 4 × 4 pixels as shown in FIG. Replace the block with the code number corresponding to the code vector. FIG.
In the example, the macro block at the right corner of a certain frame (current frame 5) is replaced with a code number “30”, and three surrounding macro blocks are replaced with a code number “15”.
It is shown that "10" and "20" have been replaced.

Such a vector quantization process is performed for each frame (previous frame 6, second-previous frame 7,...), And image data of each frame is stored.

The first embodiment relates to data compression for performing both the above-described vector quantization processing and motion vector detection processing. FIG. 1 is a block diagram illustrating a configuration of the data compression device according to the first embodiment. Hereinafter, the configuration and operation of the data compression device according to the first embodiment will be described with reference to FIG.

First, data of an original image is input from the image input unit 11. Then, the input original image data is sent to the motion vector detection unit 12 and the difference image calculation unit 14. The motion vector detection unit 12 compares the image of the current frame supplied from the image input unit 11 with the image of the previous frame temporally one before the current frame supplied from the vector quantization image reproduction unit 16. To detect a motion vector. That is, as shown in FIG.
The image of the block with the _n-1 move several places vertically and horizontally within a predetermined range, and several places of block 8 after moving (shown only one place in the drawing), a fixed block 9 in the current frame f _n A total sum D of differences between each pixel and the image is calculated. Then, the displacement amount (d of the block in the direction of the block 8 in which the difference data D is minimum)
x, dy) as the motion vector 10.

The motion vector detected by the motion vector detecting unit 12 is sent to a motion vector image reproducing unit 13 and a vector quantized image reproducing unit 16. The detected motion vector is transmitted as a part of the compressed data, for example, on a transmission path, or transmitted to a storage medium and stored. The motion vector image reproducing unit 13 predicts the image data of the current frame using the motion vector supplied from the motion vector detecting unit 12 and the image data of the previous frame, and approximately approximates the actual image data of the current frame. Generate forecast data.

The generated prediction data is sent to the difference image calculation unit 14 and the vector quantization image reproduction unit 16. In the difference image calculation unit 14, the motion vector image reproduction unit 13
The difference between the prediction data of the current frame supplied from the image input unit 11 and the image data of the actual current frame supplied from the image input unit 11 is calculated. The difference data calculated in this way is
It is sent to the difference image vector quantization unit 15. Then, the difference image vector quantization unit 15 performs vector quantization on the difference data with reference to the codebook stored in the codebook storage unit 17. In the present embodiment, each code vector stored in the codebook storage unit 17 is not a code vector for the original image itself as shown in FIG. This is a code vector having a changed pattern. The data (code number sequence) vector-quantized by the difference image vector quantization unit 15 is sent as a part of the compressed data and sent to the vector quantization image reproduction unit 16.

In the vector quantized image reproducing section 16, first, the original difference data is reproduced from the vector quantized code number sequence with reference to the code book stored in the code book storing section 17. Then, a reproduced image is generated from the difference data, the prediction data supplied from the vector image reproducing unit 13, and the motion vector supplied from the motion vector detecting unit 12. The reproduced image generated in this way is supplied to the motion vector detection unit 12 as image data of the previous frame when the motion vector detection unit 12 detects a motion vector.

As described above, according to the first embodiment of the present invention, a motion vector is obtained to generate prediction data close to the actual current frame, and the prediction data and the image data of the actual current frame are generated. By performing vector quantization on the difference image obtained from the above, data compression with a higher compression ratio can be performed without deteriorating the image quality of the reproduced image on the decompression side, compared to the case where the image data of each frame is simply vector-quantized. It is possible to do. This is because, when the above-described difference image is obtained, the luminance value can be more concentrated around “0”, so that the number of code vector patterns can be reduced and the bit length of each code number can be reduced. At the same time, the code vector pattern itself can be simplified, and each macro block can be represented by a code vector having a higher similarity.

(Second Embodiment) Next, a second embodiment of the present invention will be described with reference to FIG. Second
In the embodiment, among the code number strings obtained by the vector quantization of the difference image described in the first embodiment,
Only the code number of the part where the code vector of the macroblock corresponding to the position between the moving image frames is dissimilar is output as compressed data, and for the similar part, the code number already sent in the previous frame is decompressed. By using it sometimes, the compression rate is further improved.

FIG. 5 is a block diagram showing the configuration of the data compression device according to the second embodiment. As shown in FIG. 5, in the second embodiment, an output control processing unit 21 is added to the data compression apparatus of the first embodiment shown in FIG. I have.
In FIG. 5, the other components are the same as those in the first embodiment, and thus the same reference numerals are given and the description is omitted here.

In the second embodiment, the code number sequence vector-quantized by the differential image vector quantization unit 15 is further compressed by the output control processing unit 21.
The code number sequence output from the output control processing unit 21 and the motion vector data output from the motion vector detection unit 12 are sent as compressed data.

The output control processing section 21 compares two frames among a plurality of frames constituting a moving image in units of macroblocks, and does not output, as compressed data, a code number having a similar code vector between the two frames ( The data amount is reduced by not updating from the code number corresponding to the position in the previous frame).
Here, when the similarity is determined, the determination is made using both the difference of the total luminance value and the spatial distance (Manhattan distance) between the corresponding macroblocks of each frame.

Hereinafter, a method of determining whether or not corresponding macroblocks of adjacent frames are similar in the output control processing unit 21 will be specifically described. In the present embodiment, as described above, when determining the similarity between macroblocks, the difference between the spatial distance (Manhattan distance) of the code vector between the two macroblocks and the sum of the luminance of each pixel between the two macroblocks is determined. Two elements, the absolute value, are used as criteria.

Here, when image data is represented by a multidimensional vector, it is known that the following equation is mathematically established between the absolute value of the difference of the luminance sum and the spatial distance.

[0033]

(Equation 1)

Here, the left side indicates the spatial distance between the two vectors, and the right side indicates the difference absolute value of the luminance sum. A _i denotes a code vector of a macro block in the current frame, and B _i denotes a code vector of a macro block in the previous frame, which corresponds to A _i in position.

That is, the above equation indicates that the spatial distance between two code vectors is always equal to or greater than the difference between the luminance sum values of the two code vectors. FIG.
FIG. 4 is a schematic diagram showing a correlation between the absolute value of the difference between the luminance sums and the spatial distance. The difference between the luminance sum values and the spatial distance for the code vectors of the corresponding macroblocks in the current frame and the previous frame are obtained, and are plotted on the characteristic diagram of FIG. Is larger, so the plot position is always the area between the straight line 8 and the vertical axis.

In the characteristic shown in FIG. 7, the similarity between macroblocks increases as the distance from the origin increases, and the similarity between macroblocks decreases as the distance from the origin increases. Therefore, for the current frame, the code number obtained by vector quantization is updated from the previous frame for a macroblock plotted at a position far from the origin, and the code number is updated for a macroblock plotted at a position near the origin. Instead, it is necessary to use the code number obtained by the vector quantization of the previous frame also in the current frame.

Therefore, in the present embodiment, first, the similarity of the code vector between macroblocks is roughly determined using the difference of the luminance sum values, and macroblocks dissimilar to each other are newly added in the current frame. Update code numbers. On the other hand, for the macroblock group determined to be similar, the spatial distance of the code vector between the macroblocks is further obtained, and the similarity is determined again.

FIG. 6 first shows a rough similarity judgment based on the difference of the luminance sum values. As a result of the rough judgment based on the luminance sum values, the similarity judgment based on the spatial distance is further performed on the code vector determined not to be updated. This shows a method of deciding data to be finally updated by performing the above operation.

FIG. 6A shows the code vector (P _t (i)) of the macro block in the current frame and the code vector (P _t ) of the macro block in the previous frame.
_In the characteristic diagram, the difference absolute value of the luminance sum with _t-1 (i)) is calculated for the code vector of each macroblock, and this is plotted on the horizontal axis, and the frequency of appearance of each luminance sum difference absolute value is plotted on the vertical axis. is there. Here, the threshold value of the luminance sum for determining the similarity is, for example, “112”. in this case,
481 of all 9600 codes extracted from one frame
Since the difference in the sum of luminance values of the 0 code exceeds the threshold value, the code vectors of the macroblocks are determined to be dissimilar. Therefore, these 4810 code numbers become code numbers to be newly updated in the current frame.

Then, as shown in FIG. 6B, the code (9600-
(4810 = 4790 codes), the similarity is continuously determined based on the spatial distance. FIG. 2 (b)
For 790 macroblocks, the code vector of the macroblock in the current frame (P _t (i))
And the spatial distance between the macroblock code vector (P _t-1 (i)) in the previous frame and the code vector of each macroblock are calculated and taken on the horizontal axis, and the appearance frequency of each spatial distance value is taken on the vertical axis. FIG. In determining the similarity based on the spatial distance, the threshold of the spatial distance is set to, for example, “248”. In this case, out of the 4790 code vectors determined to be dissimilar in the determination based on the luminance sum value, 479 code vectors exceed the threshold of the spatial distance. Therefore, these 4
79 code numbers are further added as updated code numbers.

This will be described with reference to FIG.
In the determination of the similarity based on the luminance sum value shown in (a), 4810 code numbers corresponding to macroblocks having a luminance sum difference larger than the luminance sum difference threshold “112”, that is, The code number of the macroblock plotted in the area A in FIG. 5 is selected as the update data. In the determination of the similarity based on the spatial distance shown in FIG. 6B, the remaining 4790 codes that are not to be updated in the determination of the similarity based on the luminance sum value are higher than the spatial distance threshold “248”. The 479 code numbers having a large spatial distance, that is, the code numbers of the macroblocks plotted in the area B in FIG. 7 are newly added as update codes.

As a result, even if the data is not updated in the judgment based on the luminance sum value shown in FIG. 6A, a code having a large spatial distance has a code number to be newly updated in the current frame. . Therefore, the image quality of the reproduced image can be improved by adding, as a new update code, a code that is similar only by the determination based on the total luminance value.

For example, as shown in FIG. 8, in the code vectors of the macroblocks corresponding to the current frame and the previous frame, the luminance sum values are substantially equal and the difference is small, but the spatial distance may be large. (If it belongs to the area B in FIG. 7). In this case, as is apparent from FIGS. 8A and 8B, since both code vectors are dissimilar, it is necessary to output the code number obtained in the current frame as an update code. However, the code number is not updated for the macroblock belonging to the area B only by determining the similarity based on the difference in the luminance sum value. On the other hand, if the similarity is determined based on the spatial distance as in the present embodiment, the macroblocks between the current frame and the previous frame may have the same total luminance value even if they have the same total luminance value. Specifically, it can be determined as dissimilar and can be added as an update code.

Also, in the code vector of the macroblock corresponding to the current frame and the previous frame, the spatial distance may be small, but the difference in the total luminance value may be large (the space A in the area A in FIG. 7). A triangular area where the distance is less than its threshold "248"). Also in this case, not only the determination of the similarity based on the spatial distance but also the determination of the similarity based on the difference of the luminance sum values, the code vectors belonging to the area of the triangle are determined to be dissimilar, and are determined as update codes. Can be output.

As is apparent from FIG. 7, the similarity is first determined from the spatial distance based on the threshold value (248) of the spatial distance, and the code number determined to be dissimilar is calculated from the spatial distance calculation. Next, as a target, the threshold value (1
Even if the similarity is determined based on 12), as long as the threshold value of the difference between the spatial distance and the luminance sum does not change, the code number extracted as the update code does not change as a result. Therefore, the similarity may be determined first using the spatial distance, and then the similarity may be determined using the difference in the luminance sum values.

However, the luminance sum value of the code vector can be calculated in advance and tabulated, and the absolute difference value of the luminance sum value can be easily calculated using the scalar value in the table. it can. On the other hand, for spatial distance, every time vector quantization is performed for a certain frame, a complicated vector operation must be performed using the code vector obtained in the frame and the code vector obtained in the previous frame. Therefore, it is better to first make a rough judgment using the difference of the luminance sum values to extract dissimilar codes, and to reduce the number of code vectors for calculating the spatial distance. There is an advantage that the amount of calculation as a whole can be minimized as compared with the case where distance calculation is performed.

Next, the configuration and data flow of the output control processing section 21 will be described with reference to FIG. FIG.
3 is a block diagram illustrating a configuration of an output control processing unit 21. FIG.

The difference image vector quantization unit 15 shown in FIG.
The code number sequence of each macroblock in the previous frame is stored in the previous frame code number sequence storage unit 30, and the code number sequence of each macroblock in the current frame is stored in the current frame code. It is stored in the number sequence storage unit 31. Each code number is the same as the code number in the code book described in FIG. 3, and is a code number after vector quantization by the difference image vector quantization unit 15.

The code number of each macroblock in the previous frame is sent from the previous frame code number string storage section 30 to the luminance sum value / threshold value calculation processing section 32. The code number of the code vector of each macroblock in the current frame is also sent from the current frame code number string storage unit 31 to the luminance sum value / threshold value calculation processing unit 32.

The luminance sum value / threshold value calculation unit 32 refers to the corresponding code vector from the codebook storage unit 36 based on the transmitted code numbers of the macroblocks of the previous frame and the current frame, and calculates the total luminance. Is calculated, and a threshold value calculation process is performed. By this threshold value calculation process, a set of code numbers corresponding to code vectors having a difference in luminance sum value larger than a predetermined threshold value of difference in luminance sum value is selected as an update code. The update code is sent to and stored in the update data string storage unit 34.

Next, the spatial distance / threshold value calculation processing section 33
Then, the code vector determined to be similar by the luminance sum value / threshold value calculation processing unit 32 is further processed by the output control processing unit 21 and the macroblock portion having low correlation of the code vector between the two frames is processed. Only the code number is output as compressed data. Therefore, a spatial distance is calculated from the code vector data. Then, threshold processing is performed based on a predetermined threshold of the spatial distance, and an additional update code is selected. This additional update code is also sent to and stored in the update data string storage unit 34.

As described above, according to the second embodiment of the present invention, the code number sequence vector-quantized by the differential image vector quantization unit 15 is further processed by the output control processing unit 21. By doing so, only the code number of the macroblock portion having a low correlation of the code vector between two frames is output as compressed data. Therefore, the amount of compressed data can be significantly reduced as compared with the case where all the code number sequences obtained by the difference image vector quantization unit 15 are output. Moreover, in the output control processing unit 21 of the present embodiment, when determining the similarity between the code vectors of the respective macroblocks in the current frame and the previous frame, the determination based on the difference between the luminance sum values and the determination based on the spatial distance are performed. By doing so, it becomes possible to make a more accurate similarity determination as compared with a method of determining whether to update or not update the code number using only one of the determination criteria. In particular, by performing the determination based on the difference between the luminance sum values first, the amount of calculation as a whole can be reduced.

In the above embodiment, an example is described in which the difference between the luminance sum values and the Manhattan distance are used as determination criteria. However, instead of the luminance sum values, weighted luminance sum values, average luminance values, and average weighted luminance values are used. An additional value or the like may be used, or another feature amount capable of expressing a spatial distance between two vectors may be used instead of the Manhattan distance.

(Modification 1) Next, Modification 1 of the second embodiment of the present invention will be described with reference to FIG.
Here, in the data compression device of the second embodiment,
When judging the similarity of the code vectors between adjacent frames from the difference of the luminance sum values or the spatial distance, instead of setting a fixed threshold value from the outside as in the second embodiment, the similarity Is small (the difference between the luminance sum values is large,
The threshold is generated internally by externally giving the percentage of the code number to be the update code from the one with the larger spatial distance).

This will be described with reference to FIG. FIG.
In the determination of the similarity based on the difference between the luminance sum values shown in FIG.
As in the first embodiment, a fixed value “112” is given as the threshold value of the difference between the luminance sums. In this case, of all 9600 codes extracted from one frame, 48
As for the 10 codes, since the difference of the luminance sum value exceeds the threshold value, the corresponding 4810 code numbers are output as the code numbers to be newly updated in the current frame.

On the other hand, in the determination of the similarity based on the spatial distance shown in FIG. 6B, the threshold processing based on the spatial distance is further performed on the 4790 code which is not updated in the determination based on the total luminance value. In the present embodiment, the target 4
Top 10% of 790 codes in order of increasing spatial distance
The processing is performed so that the code number corresponding to the code vector is used as the update code. In this case, 479 code numbers, which are 10% of the 4790 codes, are added as code numbers to be updated, and in this case, the threshold value of the spatial distance is set to “248”. Note that this threshold value matches the fixed threshold value described in the first embodiment, but this coincides only by chance.

That is, depending on the type of moving image to be compressed, there are moving images that are sharply moving and those that are not, and the distribution shown in FIGS. 6A and 6B changes depending on the image type. Therefore, by giving, as a parameter, the ratio of what percentage is output as an update code from the top, the threshold value of the spatial distance dynamically changes according to the type of moving image. If the threshold values of the two judgment criteria are both fixed values as in the second embodiment, the data amount of the compressed data increases if the threshold value is reduced in accordance with a moving image with a lot of movement. If the threshold value is increased and the data amount is reduced, the quality of a reproduced image is deteriorated in a moving image in which movement is intense.

However, if the ratio is externally given as a parameter as in the present embodiment, an optimum threshold value suitable for the movement of the original image can be dynamically and automatically generated internally. Therefore, the code numbers of many macroblocks can be used as the update code for an image with rapid movement, and the code numbers of unnecessary macroblocks need not be updated for an image with little movement.
This makes it possible to achieve a high compression ratio while maintaining the image quality of the reproduced image.

FIG. 10 is a block diagram showing the configuration and data flow of the output control processing unit 21 according to the first modification of the second embodiment. The configuration and data flow of the output control processing unit 21 according to the first modification will be described with reference to FIG. In FIG. 10, the components having the same functions as those of the components in the output control processing unit 21 described in FIG. 9 are denoted by the same reference numerals.

The difference image vector quantization unit 15 shown in FIG.
The code number sequence of each macroblock in the previous frame is sent from the previous frame code number sequence storage unit 30 to the luminance sum value / threshold value calculation processing unit 32 among the code number sequences of the respective frames that have been vector-quantized by (1). Further, the code number of each macroblock in the current frame is sent from the current frame code number string storage section 31 to the luminance sum value / threshold value calculation processing section 32.

The luminance sum value / threshold value calculation processing unit 32 corresponds to each code number stored in the codebook data storage unit 36 based on the code numbers of the macroblocks of the previous frame and the current frame that have been sent. The difference absolute value of the luminance sum is calculated with reference to the code vector, and a threshold value calculation process is performed. By this threshold value calculation process, a set of code numbers corresponding to code vectors having a difference in luminance sum value larger than a predetermined threshold value of difference in luminance sum value is extracted as an update code. The extracted update code is sent to and stored in the update data string storage unit 34.

Next, the spatial distance calculating section 40 determines, for the code vectors determined to be similar in the determination based on the difference between the total luminance values and the total luminance value in the above-mentioned luminance total value / threshold value calculation processing section 32, the spatial distance from the code vector data. Is calculated to obtain a distribution as shown in FIG.

The spatial distance threshold value determining unit 41 determines the similarity based on the spatial distance from the spatial distance distribution of each macroblock between the current frame and the previous frame, which is obtained by the spatial distance calculating unit 40. Determine the threshold. In determining the threshold value, a predetermined percentage of the number of codes becomes a dissimilar code with respect to the total number of all the codes for which the similarity of each code vector is to be determined based on the percentage information given from the outside. The threshold is determined as follows. That is, the threshold is set so that, from the codes similar in the similarity determination based on the difference of the luminance sum values, up to a predetermined number of codes having a large spatial distance are extracted as update codes.

Then, based on the threshold value determined by the spatial distance threshold value determining section 41, the threshold value calculation processing section 42 performs threshold processing based on the spatial distance. Thereby, the code number of the macroblock determined to be dissimilar among the code vectors of the macroblock for which the spatial distance calculation has been performed is further extracted as an update code. The additional update code thus extracted is also sent to and stored in the update data string storage unit 34.

As described above, according to the first modification of the second embodiment of the present invention, the highest percentage from the smallest similarity to the total number of all the codes for which the similarity is to be determined is determined. The expressed ratio is given as a parameter of the judgment criterion. By using the ratio from the top as a parameter in this way, by setting an appropriate ratio, it is possible to update the code numbers of many macroblocks for an image with rapid movement, and for an image with little movement, Since it is not necessary to update useless macroblock code numbers, a high compression rate can be obtained while maintaining the image quality of the reproduced image.

Although the method of setting the threshold of the spatial distance based on a predetermined ratio to the total number of codes has been described with reference to FIG. 7, it is assumed that the spatial sum calculation is performed first and then the luminance sum calculation is performed. May set a predetermined ratio to the total number of codes for the difference between the luminance sum values. Further, the ratio may be given as a parameter with respect to both the criterion of the total luminance value and the spatial distance. If the threshold values of both criteria are given as a ratio, the total number of code numbers finally updated (compressed data amount) can always be constant regardless of the type of image. . On the other hand, when only one of the thresholds of the criterion is given as a ratio, the total number of update codes changes depending on the type of image. Therefore, for example, when an update code is sent on a transmission line, if only one is given at a ratio, it is necessary to allow a margin for the transmission width and increase the transmission width. There is an advantage that a minimum optimum transmission width can be set.

(Modification 2) Next, Modification 2 of the second embodiment of the present invention will be described. Modification 2 of the second embodiment is that, in the data compression device of the second embodiment, a predetermined number of code numbers are used when judging the similarity of the code vectors from the difference in luminance sum value or the spatial distance. This is output as an update code.

That is, in the second modification, when the spatial distance calculation is performed after the luminance sum calculation as shown in FIG. 6, for example, it is determined whether or not each macroblock has greatly changed between adjacent frames. When determining based on the value difference, it is determined whether or not the number of code vectors having a difference in luminance sum value exceeding the threshold value “112” of the difference in luminance sum value exceeds a predetermined number. . If it exceeds, the next spatial distance calculation is not performed. If not, the spatial distance calculation is further performed to update the code numbers of the upper 10% from the farthest one in the range not exceeding the predetermined number. Extract further as code.

For example, if the number of codes to be updated in the determination of the similarity based on the difference in the luminance sum value is set to, for example, 2000 codes, and if the number of codes extracted as dissimilar due to the difference in the luminance sum value is 2000 codes or less, Further, an update code extraction process based on the similarity determination using the spatial distance as a criterion is further performed. At this time, the number of update codes newly extracted is a number obtained by subtracting the number of codes extracted in the similarity determination based on the difference of the luminance sum value from the 2000 codes. As a result, the total number of code numbers extracted as update codes based on both the difference between the luminance sum values and the spatial distance can be set to 2000 codes.

On the other hand, when the number of update codes extracted as dissimilar due to the difference in the luminance sum exceeds 2000 codes, all code numbers of 2000 codes or more are output as update codes, and the similarity based on the spatial distance is output. Judgment of degree is not performed. Therefore, in this case, the calculation of the spatial distance can be omitted, and the processing can be simplified. Of the more than 2000 update codes extracted as dissimilar due to the difference in the luminance sum value,
Only the 000 code numbers may be output as update codes, and the total number of update codes may be set to 2000 codes.

As described above, Modification 2 of the second embodiment
According to the above, the maximum number of codes to be updated is determined in advance, so that for an image with a small amount of motion, the first and second codes described above can be used while maintaining a small number of code updates equal to or less than a predetermined number.
As in the case of the embodiment, the number of update codes can be limited by the maximum update number for an image having a large motion. Therefore, for example, when transmitting a processed code number sequence to a transmission path, it is possible to deliver a moving image with a predetermined transmission data amount or less, and to set the transmission width of the transmission path to a size corresponding to the allowable amount. It can be set.

FIG. 11 is a block diagram showing the configuration of the output control processing section 21 and the flow of data in the second modification of the second embodiment. The configuration and data flow of the output control processing unit 21 according to the second modification will be described with reference to FIG. In FIG. 11, components having the same functions as those of the components in the output control processing unit 21 described in FIG. 9 are denoted by the same reference numerals.

The difference image vector quantization unit 15 shown in FIG.
The code number sequence of each macroblock in the previous frame is sent from the previous frame code number sequence storage unit 30 to the luminance sum value / threshold value calculation processing unit 32 among the code number sequences of the respective frames that have been vector-quantized by (1). Further, the code number of each macroblock in the current frame is sent from the current frame code number string storage section 31 to the luminance sum value / threshold value calculation processing section 32.

The luminance sum value / threshold value calculation processing unit 32 corresponds to each code number stored in the code book data storage unit 36 on the basis of the code numbers of the macro blocks of the previous frame and the current frame sent. The difference absolute value of the luminance sum is calculated with reference to the code vector, and a threshold value calculation process is performed. By this threshold value calculation process, a set of code numbers corresponding to a code vector having a difference in luminance sum value larger than a predetermined threshold value of the difference absolute value difference of luminance sum is extracted as an update code. . The extracted update code is sent to and stored in the update data string storage unit 34.

Next, the update data number comparing section 50 determines whether or not the number of update codes extracted by the luminance sum value / threshold value calculation processing section 32 is equal to or greater than a predetermined number (here, 2000). I do. Then, the number of extracted update codes is 2
If the number exceeds 000, the update code is not extracted based on the similarity determination based on the spatial distance thereafter. On the other hand, if the number of update codes extracted in the similarity determination based on the total luminance value is 2000 or less, the update codes are further extracted in the similarity determination process based on the spatial distance, and the total number of update codes becomes 2,000. To do.

That is, if the number of update codes extracted by the luminance sum value / threshold value calculation processing unit 32 is 2000 or less, the spatial distance calculation unit 51 causes the luminance sum value / threshold value calculation processing unit to execute. In step 12, the spatial distance for the code vector corresponding to the code number not extracted as the update code is calculated, and the distribution as shown in FIG.

Then, the threshold value for the spatial distance is set in the spatial distance threshold value determining section 52 based on the calculation result of the spatial distance. Here, as described in the first modification,
The threshold value for the spatial distance is set to a value such that the number of codes at a predetermined ratio is updated as a dissimilar code based on the ratio information given from the outside, with respect to the total number of codes to be subjected to similarity determination. Is done.

Then, based on the threshold value determined by the spatial distance threshold value determining unit 52, the threshold value calculation processing unit 53 performs threshold processing of the code vector based on the spatial distance. As a result, the code number of the macroblock determined to be dissimilar is extracted from the code vectors of the macroblock for which the spatial distance operation has been performed. At this time, the threshold value calculation processing unit 53 extracts the code numbers determined to be dissimilar from the one with the smallest similarity in order, and sequentially supplies the code numbers to the update data number comparison unit 54.

The update data number comparing section 54 counts the sum of the code numbers sequentially supplied from the threshold value calculating section 53 and the code numbers extracted in the luminance sum value / threshold value calculating section 32. It is determined whether or not the total has reached a predetermined number (here, 2000). Then, when the sum of the code numbers exceeds a predetermined number, extraction of the update code by the threshold value calculation processing unit 53 is stopped. In this way, the additional update code extracted by the threshold value calculation processing unit 54 is also sent to the update data string storage unit 14 and stored.

As described above, according to the modified example 2 of the second embodiment, the number of updated codes can be limited by the maximum number of updated codes, so that the number of updated codes is different from that of the previous frame as in the case of an image having a large motion. Moving image data can be compressed to a predetermined data amount or less even for image data that is assumed to have a large code number updated as similar.

(Third Embodiment) Next, a third embodiment of the present invention will be described with reference to FIG. In the third embodiment, the basic operation is the same as that of the first embodiment. However, each frame constituting a moving image is classified into a key frame and a predicted frame, and a motion vector of the key frame and the predicted frame is calculated. By changing the detection method, the quality of a reproduced image can be improved.

FIG. 12 is a schematic diagram showing the arrangement of key frames and predicted frames. In FIG.
Indicates frames constituting a moving image, and the first frame, the fourth frame, the seventh frame, and the tenth frame indicate key frames. Among these key frames, the first frame is a frame serving as a reference for all frames. The second frame, the third frame, the fifth frame, the sixth frame, the eighth frame, the ninth frame, the eleventh frame, and the twelfth frame indicate predicted frames. For the first frame, which is a reference for all frames, all code numbers obtained by vector quantization of this frame are output as compressed data. That is, for the first key frame, all the obtained code numbers become update codes and are output on the transmission path or to the recording medium.

Of these frames, the fourth, seventh,
The motion vector detection and difference image generation processing in each of the tenth key frames are performed using the image data in the temporally previous key frame. For example, detection of a motion vector and generation of a difference image are performed between the fourth frame and the first frame, and detection of the motion vector and generation of the difference image between the fourth frame and the fourth frame are performed. Generation has been done. As described above, each key frame generates a difference image between the key frame and the previous key frame having a larger amount of information, and the number of frames interposed therebetween is small, so that the image quality is higher than the predicted frame. Has data.

On the other hand, in the predicted frame, the detection of a motion vector and the generation of a difference image are performed between a key frame that is temporally one before. For example, if a motion vector is always detected between the immediately preceding frame and a predicted frame, a frame to be further predicted from the predicted data occurs, and an error from the original image increases. Therefore, in the present embodiment, detection of a motion vector and the like are performed with a key frame in which the image quality is less deteriorated.

As described above, in a plurality of frames to be subjected to data compression, key frames are arranged every few frames, and a plurality of frames between the key frames are used as predicted frames, and the compressed data of the key frames is reduced to the extent that there is no problem in image quality. By making it possible to reproduce a high-quality picture with a large amount, it is possible to suppress deterioration of a reproduced image of the entire moving image.

As described above, according to the third embodiment of the present invention, each frame constituting a moving image is classified into a key frame and a predicted frame, and By performing data compression by motion vector detection and differential image vector quantization processing described in the first embodiment, it is possible to reduce the data amount of the entire moving image while maintaining the image quality of the reproduced image on the decompression side.

(Fourth Embodiment) Next, a fourth embodiment will be described with reference to FIG. In the fourth embodiment, similarly to the third embodiment, each frame constituting a moving image is classified into a key frame and a predicted frame. The difference from the third embodiment is that both the key frame and the predicted frame detect a motion vector, perform data compression by vector quantization of a difference image, and perform As described in the second embodiment, the similarity of the code vectors is determined based on the difference between the luminance sum values and the spatial distance, and only the code numbers corresponding to the code vectors determined to be dissimilar are updated. .

As described above, in a plurality of frames to be subjected to data compression, key frames are arranged every few frames, and a plurality of frames between the key frames are used as prediction frames, and the compressed data is reduced to a level where there is no problem in the image quality of the key frames. By making it possible to reproduce a high-quality picture with a large amount, it is possible to suppress deterioration of a reproduced image of the entire moving image.

According to the fourth embodiment of the present invention,
Classifying each frame constituting a moving image into a key frame and a predicted frame, and performing motion vector detection and vector quantization of a difference image described in the first embodiment between key frames and between the key frame and the predicted frame; By performing the output control process described in the second embodiment, the data amount of the entire moving image can be further reduced while maintaining the image quality of the reproduced image.

(Fifth Embodiment) Next, a fifth embodiment will be described with reference to FIG. Also in the fifth embodiment, a plurality of frames constituting a moving image are composed of key frames and predicted frames. However, in the present embodiment, the data compression method is changed between the key frame and the predicted frame.

That is, as shown in FIG. 14, in the fifth embodiment, as described in the fourth embodiment, a motion vector is detected for a key frame, a difference image is obtained, and data obtained by vector quantization is obtained. After the compression, the similarity of the code vectors is determined for each of the obtained code number sequences based on the difference between the luminance sum values and the spatial distance, and only the code numbers corresponding to the dissimilar code vectors are updated. Like that.

On the other hand, for the predicted frame, only the motion vector detection described in the first embodiment is performed, and only the obtained motion vector is output as compressed data. Thus, the data amount of the predicted frame can be reduced as compared with the key frame data, and the data amount of the entire moving image can be significantly reduced.

As described above, according to the fifth embodiment of the present invention, each frame constituting a moving image is classified into a key frame and a predicted frame.
By performing a series of processing similar to that described in the fourth embodiment, the image quality is given priority to the extent that there is no problem with the image quality in the key frame, and the image quality is prioritized. By performing only the above and outputting only the motion vector as compressed data, it is possible to prioritize the reduction of the data amount in the predicted frame.
Thus, the data amount of the entire moving image can be further reduced while maintaining the image quality of the reproduced image. Further, the processing of the prediction frame can be simplified because only the detection of the motion vector is required.

(Sixth Embodiment) Next, a sixth embodiment will be described with reference to FIG. Also in the sixth embodiment, a plurality of frames constituting a moving image are classified into a key frame and a predicted frame, and the data compression method is changed between the key frame and the predicted frame. That is, the sixth
In the third embodiment, only the motion vector of the predicted frame is output as compressed data as in the fifth embodiment. On the other hand, for the key frame, after performing normal vector quantization (vector quantization on the original image itself, not on the difference image) as shown in FIG. 3, the similarity determination described in the second embodiment is performed. And outputs only dissimilar code numbers. Therefore, motion vector detection and difference vector quantization processing are not performed between key frames.

As described above, according to the sixth embodiment of the present invention, each frame constituting a moving image is classified into a key frame and a predicted frame, and a normal vector quantization is performed on the key frame. After that, by performing a series of processing similar to the output control processing described in the second embodiment, it is possible to give priority to image quality by giving a key frame an amount of information to the extent that there is no problem in image quality. For a frame, only detection of a motion vector is performed and only the motion vector is output as compressed data, so that reduction of the data amount can be prioritized in a predicted frame. Thus, the data amount of the entire moving image can be further reduced while maintaining the image quality of the reproduced image. Also,
As for the processing of the predicted frame, only the detection of the motion vector is sufficient, so that the processing can be simplified.

(Seventh Embodiment) Next, a seventh embodiment will be described with reference to FIG. In the seventh embodiment, a plurality of frames constituting a moving image are separated into a key frame and a predicted frame, and different processes are performed for a luminance signal and a chrominance signal. That is, as shown in FIG. 16A, the same processing as in the sixth embodiment is performed on the luminance signal (Y signal). That is, for the key frames, data compression is performed by the output control calculation process described in the second embodiment after normal vector quantization, and for the predicted frame, only the motion vector is output to perform data compression. Further, the same processing as in the third to fifth embodiments may be performed on this key frame.

On the other hand, for the color signal (U signal), no distinction is made between a key frame and a predicted frame, and normal vector quantization (see FIG. 3) is performed for each of a plurality of frames, and then the second embodiment is performed. Data compression based on the similarity determination described in the embodiment is performed between adjacent frames. The rates of the luminance signal and the chrominance signal are represented by Y: U: V = 4: 1:
In the case where the color signal is decimated and vector quantized as in the case of 1, for the decimated color signal, a difference image is obtained from the motion vector and vector quantization is performed. When they are noticeable. Also, due to the nature of the image, the data amount of the color signal can be sufficiently reduced without affecting the image quality by simply performing output control processing after normal vector quantization.

As described above, according to the seventh embodiment of the present invention, image data is separated into a luminance signal and a chrominance signal, and each frame constituting a moving image is classified into a key frame and a predicted frame. For the luminance signal only, processing is performed by changing the compression method between the key frame and the predicted frame. As a result, it is possible to increase the data compression ratio of the entire moving image while maintaining the image quality of the reproduced image.

As for the luminance signal, between key frames, each frame is vector-quantized,
The data compression of the entire moving image can be achieved by performing data compression by the ABS processing described in the second embodiment, and detecting only the motion vector detection to generate prediction data between the key frame and the prediction frame. It becomes possible.

(Eighth Embodiment) Next, an eighth embodiment of the present invention will be described with reference to FIG. In the eighth embodiment, the key frames are the first frame and the seventh frame, and the second and third frames of the second to sixth frames that are predicted frames are temporally earlier than the first and seventh frames. Is compressed with the first frame, and the fifth and sixth frames are compressed with the seventh frame that is later in time. Then, for the fourth frame in the middle, data is compressed using both the first and seventh frames.

Here, the compression of the data of the second to sixth frames based on the first and seventh frames is performed by detecting a motion vector and outputting the motion vector. Further, a difference image may be obtained from the motion vector, and vector quantization may be performed on the difference image, or a compression process based on similarity determination may be further performed on the code number sequence after the vector quantization. You may.

When the image of the fourth frame, which is the predicted frame, is reproduced from the data thus compressed, the image reproduced from the first frame by using the motion vector and the seventh frame are reproduced. By calculating the average value between the corresponding pixels of the image reproduced using the motion vector and the original image, the original image can be reproduced.

As described above, a motion vector can be obtained from a future image, and a single image can be reproduced by obtaining a motion vector from two or more images.

(Ninth Embodiment) Next, a ninth embodiment of the present invention will be described. In the eighth embodiment, the second
Similarly to the second modification of the embodiment, a predetermined number of code numbers are output as update codes when judging the similarity of the code vectors from the difference in the total luminance value or the spatial distance. .

In the second modification of the second embodiment, when the number of update codes is limited, only the threshold value of the spatial distance calculation is changed. However, in the ninth embodiment, first, the first Modified example 2 of the second embodiment in that the similarity of the code vectors is determined from the difference in the sum of luminance values or the spatial distances, and the number of update codes is determined after identifying the update codes in the same manner as in the embodiment. Is different from

FIG. 18 is a block diagram showing the configuration and data flow of the output control processing section 21 in the ninth embodiment. The configuration and data flow of the output control processing unit 21 according to the ninth embodiment will be described with reference to FIG. In FIG. 18, components having the same functions as those of the components in the output control processing unit 21 described in FIG. 9 are denoted by the same reference numerals.

Next, the configuration and data flow of the output control processing section 21 will be described with reference to FIG. FIG. 18 is a block diagram illustrating a configuration of the output control processing unit 21.

The difference image vector quantization unit 15 shown in FIG.
The code number sequence of each macroblock in the previous frame is stored in the previous frame code number sequence storage unit 30, and the code number sequence of each macroblock in the current frame is stored in the current frame code. It is stored in the number sequence storage unit 31. Each code number is the same as the code number in the code book described in FIG. 3, and is a code number after vector quantization by the difference image vector quantization unit 15.

The code number of each macroblock in the previous frame is sent from the previous frame code number string storage section 30 to the luminance sum value / threshold value calculation processing section 32. The code number of the code vector of each macroblock in the current frame is also sent from the current frame code number string storage unit 31 to the luminance sum value / threshold value calculation processing unit 32.

The luminance sum value / threshold value calculation processing unit 32 refers to the corresponding code vector from the codebook storage unit 36 based on the transmitted code numbers of the macroblocks of the previous frame and the current frame, and calculates the luminance sum. Is calculated, and a threshold value calculation process is performed. By this threshold value calculation process, a set of code numbers corresponding to code vectors having a difference in luminance sum value larger than a predetermined threshold value of difference in luminance sum value is selected as an update code. The update code is sent to and stored in the update data string storage unit 34.

Next, the spatial distance / threshold value calculation processing unit 33
Then, the code vector determined to be similar by the luminance sum value / threshold value calculation processing unit 32 is further processed by the output control processing unit 21 and the macroblock portion having low correlation of the code vector between the two frames is processed. Only the code number is output as compressed data. Therefore, a spatial distance is calculated from the code vector data. Then, threshold processing is performed based on a predetermined threshold of the spatial distance, and an additional update code is selected. This additional update code is also sent to and stored in the update data string storage unit 34.

The update code stored in the update data string storage unit 34 is sent to the update data number comparison unit 61.
Then, the update code is sent from the update data number comparison unit 61 to the generated data storage unit 62. Update data number comparison unit 61
From the update data string storage unit 34 to the generated data storage unit 62
When the number of update data reaches a predetermined number (for example, 2000 codes), the number of update data transmitted from the update data string storage unit 34 to the generated data storage unit 62 is counted. Stop sending.

Here, it is desirable that the update codes sent from the update data string storage unit 34 to the generated data storage unit 62 be transmitted in order from a dissimilar code. Therefore, the update data number comparison unit 61 determines the similarity of the update codes, and transmits the codes to the generated data storage unit 62 sequentially from the code with the highest degree of dissimilarity.

FIG. 19 is a diagram schematically showing update codes sequentially sent from the update data number comparison unit 61 to the generated data storage unit 62. Regarding the transmission order of the update codes, the similarity is determined based on both the difference between the luminance sums and the spatial distance, and the update codes are sequentially generated from the update code having both the luminance difference and the spatial distance larger than the solid line 63 in FIG. The data is transmitted to the data storage unit 62. Here, the solid line 63 indicates a boundary where the transmission code exceeds a predetermined threshold value (here, 2000 codes). Therefore, by transmitting all the update codes having both the difference in luminance sum and the spatial distance larger than the solid line 63, only the desired number of update codes is stored in the generated data storage unit 6.
2 can be sent.

The position of the solid line 63 can be changed according to the number of update codes. For example, if the number of update codes is changed to the position indicated by the one-dot chain line 64, the number of update codes can be reduced to, for example, 1500 codes from the case of the solid line 63.

As described above, according to the ninth embodiment of the present invention, after extracting an update code by the method of the first embodiment, when transmitting the update code, the number of update codes is set to a predetermined value. Since the setting is performed, only the dissimilar code can be set without fail.

Therefore, in particular, when transmitting moving image transmission data, the transmission amount is limited by the transmission signal line.
By setting the number of update codes in advance by the update data number comparison unit 61, it is possible to transmit the number of update codes according to the transmission signal line.

The present embodiment is particularly suitable for data transmission of moving pictures whose transmission amount is limited, but can also be applied to still pictures. Also, the input image (previous frame code number, current frame code number) in FIG. 18 is not limited to a motion vector image, but is a difference image between a plurality of frames constituting a moving image, an original image and an image obtained by vector quantization. Can be used.

(Other Embodiments) Each functional block and processing procedure shown in the above various embodiments may be constituted by hardware, or may be constituted by CPU, MPU, R
It may be configured by a microcomputer system including an OM and a RAM, and its operation may be realized according to a work program stored in a ROM or a RAM. The present invention also includes a software program for realizing the functions described above, which is provided to a RAM so as to realize the functions of the respective functional blocks, and which is executed by operating the functional blocks according to the programs. include.

In this case, the software program itself realizes the functions of the above-described embodiments, and the program itself and means for supplying the program to a computer, for example, a recording medium storing the program Constitute the present invention. As a storage medium for storing such a program, the above-described ROM or RAM
Besides, for example, floppy disk, hard disk,
Optical disk, magneto-optical disk, CD-ROM, CD-
I, CD-R, CD-RW, DVD, zip, magnetic tape, nonvolatile memory card, or the like can be used.

When the computer executes the supplied program, not only the functions of the above-described embodiments are realized, but also the OS (operating system) or other application software running on the computer. Needless to say, such a program is also included in the embodiment of the present invention when the functions of the above-described embodiment are realized in cooperation with the above.

Further, after the supplied program is stored in the memory provided on the function expansion board of the computer or the function expansion unit connected to the computer, the program is provided on the function expansion board or the function expansion unit based on the instruction of the program. It is needless to say that the present invention also includes a case where the CPU or the like performs part or all of the actual processing, and the processing realizes the functions of the above-described embodiments.

[0123]

According to the first aspect of the present invention,
By generating a predicted image by obtaining a motion vector and performing vector quantization on a difference between the predicted image and the image of the current frame, further compression is performed between frames constituting a moving image while maintaining the image quality of each frame. It is possible to achieve an improvement in the rate.

According to the invention of claim 6 of the present invention, the key frame is processed by the vector quantizing means, the luminance sum calculating means, the spatial distance calculating means and the output control means, so that the dissimilar code is obtained. By outputting only the columns, the compression rate can be improved while maintaining the image quality of each key frame. Further, by performing only the processing by the motion vector detecting means on the predicted frame, the amount of information can be reduced as compared with the key frame. Then, by arranging the key frames with the predicted frame interposed therebetween, it is possible to further improve the compression ratio while maintaining the image quality of the entire moving image.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration example of a data compression device according to a first embodiment of the present invention.

FIG. 2 is a schematic diagram illustrating a method for detecting a motion vector.

FIG. 3 is a schematic diagram for explaining image compression by general vector quantization.

FIG. 4 is a schematic diagram for explaining data compression of a moving image using vector quantization.

FIG. 5 is a block diagram illustrating a configuration example of a data compression device according to a second embodiment of the present invention.

FIG. 6 is a schematic diagram for explaining an operation relating to a total luminance value and a spatial distance performed in a second embodiment of the present invention.

FIG. 7 is a schematic diagram showing a distribution based on a difference between luminance sum values and a difference in spatial distance according to the second embodiment of the present invention.

FIG. 8 is a schematic diagram illustrating an example of a code vector having the same total luminance value but a significantly different spatial distance.

FIG. 9 is a block diagram illustrating a configuration of an output control processing unit according to a second embodiment of the present invention.

FIG. 10 is a block diagram illustrating a configuration of an output control processing unit according to a first modification of the second embodiment of the present invention.

FIG. 11 is a block diagram illustrating a configuration of an output control processing unit according to a second modification of the second embodiment of the present invention.

FIG. 12 is a schematic diagram for explaining a data compression method according to a third embodiment of the present invention.

FIG. 13 is a schematic diagram for explaining a data compression method according to a fourth embodiment of the present invention.

FIG. 14 is a schematic diagram for explaining a data compression method according to a fifth embodiment of the present invention.

FIG. 15 is a schematic diagram for explaining a data compression method according to a sixth embodiment of the present invention.

FIG. 16 is a schematic diagram for explaining a data compression method according to a seventh embodiment of the present invention.

FIG. 17 is a schematic diagram for explaining a data compression method according to an eighth embodiment of the present invention.

FIG. 18 is a block diagram illustrating a configuration of an output control processing unit according to a ninth embodiment of the present invention.

FIG. 19 is a schematic diagram illustrating a similarity determination method before transmission according to a ninth embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Original image 2 Input image block 3 Codebook 4 Reproduction image 5 Current frame 6 Previous frame 7 Before frame 8, 9 block 10 Motion vector 11 Image input unit 12 Motion vector detection unit 13 Motion vector image reproduction unit 14 Difference image calculation unit Reference Signs List 15 Difference image vector quantization unit 16 Vector quantization image reproduction unit 17 Codebook storage unit 21 Output control processing unit 30 Previous frame code number sequence storage unit 31 Current frame code number sequence storage unit 32 Luminance sum value / threshold value calculation process Unit 33 Spatial distance / threshold processing calculation unit 34 Update data string storage unit 36 Codebook storage unit 40 Spatial distance calculation unit 41 Spatial distance threshold determination unit 42 Threshold calculation processing unit 50, 54 Update data number comparison unit 51 Spatial distance calculation unit 52 Threshold value determination unit for space distance 53 Threshold calculation processing unit 6 Number updating data comparator 62 generates data storage unit

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Takahiro Nakayama Aoba, Aoba-ku, Aoba-ku, Sendai City, Miyagi Prefecture (no address) Inside Tohoku University (72) Inventor Masahiro Ekina Aoba, Aoba-ku, Aoba-ku, Sendai City, Miyagi Prefecture (no address) Inside Tohoku University (72) Inventor Hirotsugu Iemura 4-1-1 Hongo, Bunkyo-ku, Tokyo Cosmos Hongo Building I & F Corporation F-term (reference) 5C059 KK00 KK06 MA05 MC18 MD04 MD07 MD10 NN29 PP04 PP16 RC11 RC16 TB00 TB04 TB07 TC00 TC12 TD03 TD05 TD12

Claims (26)

[Claims] 1. A motion vector detecting means for detecting a motion vector using a plurality of frames constituting a moving image, and a predicting means for generating a predicted image of a current frame using the motion vector detected by the motion vector detecting means. A difference image generating means for calculating a difference between the predicted image generated by the predicting means and the image of the input frame to generate a difference image; and a vector image generating the difference image generated by the difference image generating means. And a difference image vector quantizing means. 2. The method of claim 1, further comprising comparing code vectors of macroblocks corresponding to each other in position between two frames and calculating an absolute value of a difference in luminance sum of the two code vectors. A luminance sum value calculating means for judging code vectors larger than the threshold value as dissimilar to each other, and comparing the code vectors of the macroblocks corresponding to each other in position between the two frames, And a spatial distance calculating means for judging that code vectors having the spatial distance larger than a second threshold value are dissimilar to each other, and for the code sequence of each frame output from the vector quantizing means, Of the code vector determined to be dissimilar by the value calculating means and the code vector determined to be dissimilar by the spatial distance calculating means. Data compression apparatus according to claim 1, characterized in that an output control means for outputting only the code string corresponding to the person. 3. Each frame constituting the moving image includes a predicted frame and a key frame arranged with at least one of the predicted frames interposed therebetween. When the current frame is the key frame, the key frame includes The motion vector detecting means, the predicting means, the difference image generating means and the difference image vector quantizing means perform processing using the temporally preceding or following other key frame, and the current frame is the predicted frame. In the case of, the motion vector detecting means, the predicting means, the difference image generating means, and the difference image vector quantizing means perform processing by using the predicted frame and a temporally preceding or succeeding key frame. The data compression device according to claim 1, wherein 4. Each of the frames constituting the moving image includes a predicted frame and a key frame arranged with at least one of the predicted frames interposed therebetween, and when the current frame is the key frame, the key frame includes The motion vector detection means, the prediction means, the difference image generation means, the difference image vector quantization means, the luminance sum value calculation means, the spatial distance calculation using another key frame before or after the time. Means and the output control means, and when the current frame is the predicted frame, the motion vector detection means, the prediction means, and the difference using the predicted frame and a temporally previous or subsequent key frame. An image generation unit, the difference image vector quantization unit, the luminance sum value calculation unit, the spatial distance calculation unit, and the output control unit 3. The data compression apparatus according to claim 2, wherein the processing is performed by control means. 5. Each frame constituting the moving image includes a predicted frame and a key frame arranged with at least one of the predicted frames interposed therebetween, and when the current frame is the key frame, the key frame is The motion vector detection means, the prediction means, the difference image generation means, the difference image vector quantization means, the luminance sum value calculation means, the spatial distance calculation using another key frame before or after the time. Means for performing processing by said means and said output control means, and when said current frame is said predicted frame, performing only processing by said motion vector detection means and outputting only the obtained motion vector as compressed data. The data compression device according to claim 2. 6. A motion vector detecting means for detecting a motion vector using a plurality of frames constituting a moving image, and performing a vector quantization on each of the plurality of frames constituting the moving image, and Vector quantization means for outputting a code string obtained for each frame, and comparing the code vectors of the positionally corresponding macroblocks between the two frames, calculating the absolute difference of the luminance sum of the two code vectors, A luminance sum value calculating means for judging that code vectors having a difference absolute value of the luminance sum greater than a first threshold value are dissimilar to each other, and code vectors of macroblocks corresponding in position between the two frames. To calculate the spatial distance between the two code vectors, wherein the spatial distance is larger than the second threshold value. A spatial distance calculating means for determining that the two are not similar; a code vector determined to be dissimilar by the luminance sum value calculating means for the code string of each frame output from the vector quantization means; and the spatial distance calculating means. And output control means for outputting only a code string corresponding to both of the code vectors determined to be dissimilar to each other, wherein each frame constituting the moving image has a predicted frame and at least one of the predicted frames interposed therebetween. A key frame to be arranged, and when the current frame is the key frame, the vector quantization means using the key frame and another key frame before or after the key frame,
The luminance sum calculating means, the spatial distance calculating means, and the output control means perform the processing, and when the current frame is the predicted frame, the processing is performed only by the motion vector detecting means, and only the obtained motion vector is obtained. A data compression device that outputs the data as compressed data. 7. The image data of each frame constituting the moving image includes a luminance signal and a chrominance signal, and the separation between the key frame and the prediction frame is performed only for the luminance signal. 7. The data compression device according to any one of 6. 8. A threshold setting means for setting the first threshold value for the luminance sum value calculation means and the second threshold value for the spatial distance calculation means, Setting, for at least one of the first threshold value and the second threshold value, a ratio of the number of output codes to the total number of code vectors to be subjected to similarity determination. 7. The data compression device according to 2 or 6. 9. A total number of at least one of a code string corresponding to a code vector determined to be dissimilar by the luminance sum calculating means and a code string corresponding to a code vector determined to be dissimilar by the spatial distance calculating means. Is the third
A first code number comparing means for stopping the output processing of the dissimilar code by the luminance sum calculating means and the spatial distance calculating means when the threshold value is exceeded. Or the data compression device according to 6. 10. The total number of code strings corresponding to the code vectors determined to be dissimilar by the luminance sum calculating means and the code strings corresponding to the code vectors determined to be dissimilar by the spatial distance calculating means is a fourth. A second code number comparing means for stopping the output processing of the dissimilar code by the luminance sum calculating means and the spatial distance calculating means when the threshold value is exceeded. Or the data compression device according to 6. 11. When the total number of code strings corresponding to code vectors determined to be dissimilar by the luminance sum value calculation means exceeds the third threshold value, the calculation by the spatial distance calculation means is performed. 10. The data compression device according to claim 9, wherein the data compression device is not provided. 12. The third code number comparison means, further comprising: third code number comparison means for counting the total number of codes in the code string output from the output control means and comparing the counted number with a predetermined value. When the total number of codes reaches the predetermined value, the output from the output control means is stopped, and the number of update codes output from the output control means is limited to a predetermined value or less. 7. The data compression device according to claim 2, wherein the data compression is performed. 13. A first step of detecting a motion vector using a plurality of frames constituting a moving image, and a second step of generating a predicted image of a current frame using the motion vector detected in the first step. And a third step of calculating a difference between the predicted image generated in the second step and the image of the input frame to generate a difference image; and a difference generated in the third step. And a fourth step of vector-quantizing the image. 14. A method of calculating a difference absolute value of a luminance sum of two code vectors by comparing code vectors of macroblocks corresponding in position between two frames,
A fifth step of judging that code vectors having a difference absolute value of the luminance sum larger than a first threshold value are dissimilar to each other; and, between the two frames, code vectors of positionally corresponding macroblocks. A sixth step of calculating a spatial distance between both code vectors in comparison with each other and judging code vectors whose spatial distance is larger than a second threshold value to be dissimilar to each other; A seventh step of outputting only a code string corresponding to both the code vector determined to be dissimilar in the fifth step and the code vector determined to be dissimilar in the sixth step for the code string of the frame; 14. The data compression method according to claim 13, comprising: 15. Classifying each frame constituting the moving image into a predicted frame and a key frame arranged with at least one of the predicted frames interposed therebetween, and, when the current frame is the key frame, the key frame And the other key frame before or after in time.
And the processing of the second step, the third step, and the fourth step are performed, and when the current frame is the predicted frame, the current frame is compared with the temporally preceding or succeeding key frame. Using the first
14. The data compression method according to claim 13, wherein the processes of the second step, the third step, and the fourth step are performed. 16. Classifying each frame constituting the moving image into a predicted frame and a key frame arranged with at least one of the predicted frames interposed therebetween, and, when the current frame is the key frame, the key frame And the other key frame before or after in time.
Performing the processes of the first to seventh steps, and when the current frame is the predicted frame, performing the processes of the first to seventh steps using the predicted frame and a temporally preceding or succeeding key frame. 15. The method according to claim 14, wherein
2. The data compression method according to 1. 17. Classifying each frame constituting the moving image into a predicted frame and a key frame arranged with at least one of the predicted frames interposed therebetween, and when the current frame is the key frame, the key frame is And the other key frame before or after in time.
The processing according to the seventh to seventh steps is performed, and when the current frame is the predicted frame, only the processing in the first step is performed, and only the obtained motion vector is output as compressed data. Item 15. The data compression method according to Item 14. 18. A first step of detecting a motion vector using a plurality of frames constituting a moving image, and performing a vector quantization on each of the plurality of frames constituting the moving image to obtain each macro block in the frame. A second step of outputting a code string obtained every time, and comparing the code vectors of the positionally corresponding macroblocks between the two frames to calculate the absolute difference of the luminance sum of both code vectors, A third step of judging that code vectors having a difference absolute value of the luminance sum greater than a first threshold value are dissimilar to each other; and, between the two frames, code vectors of macroblocks corresponding in position. Computes the spatial distance of both code vectors in comparison, and dissimilars code vectors whose spatial distance is greater than a second threshold. And a code vector determined to be dissimilar in the third step and a dissimilar determination in the fourth step for the code string of each frame output in the second step. A fifth step of outputting only a code string corresponding to both of the generated code vectors, wherein each frame constituting the moving image includes a predicted frame and a key arranged to sandwich at least one of the predicted frames. When the current frame is the key frame, the second step, the third step, and the fourth step are performed using the key frame and another key frame that is temporally earlier or later. Performing the processing in the step and the fifth step, and when the current frame is the predicted frame, performing only the processing in the first step, Data compression method and outputs only the motion vectors which is as compressed data. 19. The image data of each frame constituting the moving picture includes a luminance signal and a chrominance signal, and the separation between the key frame and the prediction frame is performed only on the luminance signal. 19. The data compression method according to any one of 18. 20. The method according to claim 19, wherein the first step corresponds to the fifth step.
And the second threshold value for the sixth step are set, and at least one of the first threshold value and the second threshold value is subjected to similarity determination. 15. The data compression method according to claim 14, wherein a ratio of the number of output codes to the total number of code vectors is set. 21. The total number of at least one of a code string corresponding to a code vector determined to be dissimilar in the fifth step and a code string corresponding to a code vector determined to be dissimilar in the sixth step is 15. The data compression method according to claim 14, wherein the output processing of the dissimilar code in the fifth step and the sixth step is stopped when the third threshold value is exceeded. 22. If the total number of code strings corresponding to code vectors determined to be dissimilar in the fifth step exceeds the third threshold value, the calculation in the third step is not performed. 22. The data compression method according to claim 21, wherein: 23. The total number of code strings corresponding to code vectors determined to be dissimilar in the fifth step and code strings corresponding to code vectors determined to be dissimilar in the sixth step is the fourth. 15. The data compression method according to claim 14, wherein the output processing of the dissimilar code in the fifth step and the sixth step is stopped when the threshold value is exceeded. 24. An eighth step of comparing the total number of code strings output in the seventh step with a fifth threshold value, wherein the total number of code strings is equal to the fifth threshold value. 15. The data compression method according to claim 14, wherein the output of the code string in the seventh step is stopped when the number exceeds the threshold value. 25. A computer-readable storage medium storing a program for causing a computer to function as each unit of the data compression apparatus according to claim 1. Description: 26. A computer-readable storage medium storing a program for causing a computer to execute the procedure of the data compression method according to claim 13.