JPH02226984A - Picture data processor - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application! l'f) This invention compresses the amount of data when recording or transmitting image data, and also compresses the compressed data during playback. This invention relates to an improvement in an image data processing device that obtains original image data by decompression.

(Prior Art) As is well known, the amount of image information is enormous, and when recording or transmitting it, some kind of data compression is generally applied, and the compressed data is decompressed during playback. I am trying to restore the original image data.

FIG. 1O shows such conventional compression of image data.

The expansion means is conceptually shown. In other words, Fig. 10 (
As shown in a), a plurality of original images 11 (in the illustrated case, two
12°12. ......, each sub-block 12.12. The data amount is compressed independently for each...

Then, during playback, each subblock is 12.12°...
The data is expanded independently for each block, and as shown in Figure 1O (b), the original sub-block 12.12°.
The same sub-blocks 12', 12',
. . . are reproduced to obtain a reproduced image 11' that is the same as the original original image 11.

Here, the data compression method is the DPCM method.

Various methods have been proposed, such as an orthogonal transform method and a vector quantization method, but they are effective when used in combination with the block division type compression method as described above.

However, when performing compression and decompression processing as described above,
In conventional image data processing devices, no matter which compression method is used, compression and decompression processing is performed independently for each sub-block 12.12°. If there is an error in the compression and decompression processing of..., each sub-block 12' on the reproduced image 11'
, 12', . . . , a problem arises in that so-called block distortion occurs in which the boundaries between the lines are conspicuous in a lattice pattern.

Therefore, in the past, for example, for the peripheral pixels of each sub-block 12', 12', . .12. . . . or a means for adding white noise to the peripheral portions of each sub-block 12', 12', . . . of the reproduced image 11' is provided. However, these methods are still not able to sufficiently remove block distortion, and "blur" occurs only in the periphery of sub-blocks, which actually deteriorates the image quality. This has caused the inconvenience of inviting

(Problems to be Solved by the Invention) As described above, in the conventional image data processing device, an original image is divided into a plurality of subblocks, and each subblock is compressed and expanded independently. There is a problem in that block distortion occurs in the reproduced image.

Therefore, this invention was made in consideration of the above circumstances, and it is possible to prevent block distortion from occurring in the reproduced image even if the original image is divided into multiple blocks and each block is compressed and expanded independently. An object of the present invention is to provide an extremely good image data processing device that can perform the following operations.

[Structure of the Invention] (Means for Solving the Problems) An image data processing device according to the present invention divides an original image into a plurality of blocks, compresses the amount of data for each block, and compresses the amount of data for each block. The target image is one that decompresses each block and combines the decompressed blocks to restore the original image.

A dividing means divides the original image into a plurality of blocks such that their peripheral parts overlap, a compression means independently compresses the amount of data for each block divided by the dividing means, and an output of the compression means. A decompression means that decompresses each block independently and weights and adds the overlapping blocks in the overlapping portion of each block, and a composition means that combines each block decompressed by the decompression means to obtain the original image. It is equipped with the following.

(Function) According to the above configuration, when dividing an original image into multiple blocks, the periphery of each block overlaps with the adjacent block, and during decompression, the overlapping part of each block is Since predetermined data is weighted and added to the blocks in the reproduced original image, the pixels in the periphery of each block in the reproduced original image are generated based on the data of multiple overlapping blocks. The overlapping portions make the seams between the blocks smooth, and no block distortion occurs in the reproduced image.

(Example) Hereinafter, an example of the present invention will be described in detail with reference to the drawings. In FIG. 2, original image data supplied to an input terminal 13 is divided by a subblock division circuit 14 into square subblocks each consisting of a plurality of N×N pixels. In this case, the sub-block dividing circuit 14 operates to divide the original image data so that the peripheral parts of each sub-block have a predetermined overlap C with each other, as shown in FIG.

That is, if we pay attention to one sub-block 15 in FIG. 1, there are sub-blocks 16 .
There are overlapping parts 15a and 15b with 17, and
At the upper and lower peripheral parts of the sub-block 15 in the figure, the sub-block 1 adjacent to the upper and lower sides of the sub-block 15 in the figure
There are overlapping parts 15c and 15d with 8.19, and further, in the periphery of the four corners of the sub-block 15, there are sub-blocks 20 adjacent to the four corners of the sub-block 15.
There will be overlapped portions 15e-15h with 23.

The data of each sub-block divided by the sub-block dividing circuit 14 is subjected to data amount compression for each sub-block independently by a compression circuit 24. As mentioned above, DPC is used to compress the amount of data.
Various methods are used, such as the M method, orthogonal transform method, and vector quantization method.

The image data compressed in this way is
After being recorded on a predetermined recording medium or transmitted, it is supplied to an expansion circuit 25 on the reproduction side. This decompression circuit 25 converts the input compressed data of each sub-block into
Decompression processing is performed independently using a method corresponding to the compression method, and for overlapping parts around the periphery of each sub-block, predetermined data is weighted and added to all overlapping sub-blocks. works.

That is, the decompression circuit 25 weights and adds 1/2 to the data of each sub-block in a portion where two sub-blocks overlap, such as the overlapping portions 15a to 15d shown in FIG. and overlapping part 15e~
15h, where four subblocks overlap, 1/4 of the data in each subblock is added.
Functions to weight and add degrees.

The image data expanded for each sub-block by the expansion circuit 25 is combined by a sub-block synthesis circuit 26 and restored to the original original image data, which is taken out from an output terminal 27 and used for image display.

According to the configuration of the above embodiment, when dividing the original image data into a plurality of subblocks, the peripheral part of each subblock overlaps with the adjacent subblock, and when decompressing, the overlapping part of each subblock is Since predetermined data is weighted and added to all of the overlapping sub-blocks, the pixels in the periphery of each sub-block of the reproduced original image data are the same as those of the multiple overlapping sub-blocks. It is generated based on the data, and the seams between the sub-blocks are smoothed due to the overlapping parts, so that no block distortion can be seen in the reproduced image.

In this case, the amount of information increases by the overlapped part, but by dividing it into multiple bands and eliminating the overlapped part in the high range,
It is possible to suppress the increase in the amount of information. Further, when only the DC component has an overlapping portion, there is no increase in the amount of information, and there is no distortion in the DC component where block distortion is visually the most noticeable, so even more effects can be obtained.

Furthermore, the method of weighting the overlapping portion may be as shown in FIG. 3. That is, as shown in FIG. 3(a), sub-block 15. 17, the same figure (b) corresponds to sub-block 15.17.

As shown in (C), the non-overlapping portions are given a weight of 1, and the overlapping portions 15b are given weights that are far from the center of the block (the further away from the center of the block, the smaller the weight). Be effective.

Here, in Fig. 1, the subblocks are divided into squares of N×N pixels, but this makes calculations easier when, for example, orthogonal transformation is used as the compression method. This is because it is convenient when using arithmetic algorithms.

However, it goes without saying that the shape of the sub-blocks to be divided is not limited to squares, but may also be rectangles, triangles and circles, etc. FIG. 4 shows a case where the shape of the sub-block is circular. In this case, in the periphery of one circular sub-block 28, there are an overlapping portion 28a where two circular sub-blocks overlap and an overlapping portion 28b where three circular sub-blocks overlap.

During decompression, 1/2 is weighted and added to the data of each circular sub-block in a portion 28a where two circular sub-blocks overlap, and in a portion 28b where three circular sub-blocks overlap, , by weighting and adding 1/3 to the data of each circular sub-block, a reproduced image without block distortion can be obtained.

Since a general image has a property in which portions with equal statistical properties exist concentrically (isotropic property), a natural reproduced image can be obtained by selecting a circular shape as the sub-block shape.

Further, as shown in FIG. 4, if the circular sub-blocks are arranged so that the odd-numbered block lines and the even-numbered block lines have different phases, block distortion can be further effectively reduced.

Note that when performing 4-small amount compression on a circular sub-block, there will be some restrictions due to its shape, but for example, starting from the center pixel of the circular sub-block and scanning in a spiral pattern, DPCM compression, which encodes the difference between

By compressing the amount of image information in the manner described above, it is possible to eliminate block distortion and obtain a high-quality reproduced image. By the way, in the case of the embodiment shown in FIG. 1, data is transmitted multiple times for the overlapping portion of each sub-block, so that data compression efficiency decreases. On the other hand, in order to further reduce block distortion, it is sufficient to increase the area of the overlapping portion, but this results in a contradiction in that the amount of data increases.

FIG. 5 shows a second embodiment of the present invention, which aims to reduce the above-mentioned contradiction between image quality and amount of information. It was made by using That is, input terminal 2
The original image data supplied to filter circuit 9 is transmitted to filter circuits 30 to 3.
Input to 2 and low range.

It is separated into mid-range and high-range components.

The characteristics of each of these filter circuits 30 to 32 are as shown in 1 in FIG.
As shown in ~ ■, the combined gain for each frequency is set to 1'''. In other words, 3
The output sums of the two filter circuits 30 to 32 are separated so as to be exactly equal to the input image data.

The image data band-limited by each of the filter circuits 29 to 30 is then processed by sub-block dividing circuits 33 to 30.
35, and is divided into a plurality of sub-blocks such that their peripheral parts overlap each other with a predetermined width C. In this case, the width of the overlapping portion of each subblock is made different for each band.

For example, if the width of the overlapping part of the sub-block dividing circuit 33 for low frequency signals is "C" as shown in FIG.
/2'', and the high-frequency signal sub-block dividing circuit 35 divides the signal into sub-blocks without having any overlapping parts.

In step 1-, the data output from each sub-block dividing circuit 33 to 35 is supplied to compression circuits 36 to 38, respectively, and the data amount is compressed independently for each sub-block. The images are synchronized and output as one image data.

The image data compressed in this way is
After being recorded on a predetermined recording medium or transmitted, it is supplied to a separation circuit 40 on the reproduction side. The separation circuit 40 separates the input image data into three band components, low, middle, and high, as synchronized by the synchronization circuit 39.

The image data of each of the low, middle and high band components separated by the separation circuit 40 is then sent to an expansion circuit 4, respectively.
1 to 43. These expansion circuits 41 to 43 are
The input compressed data for each sub-block is independently decompressed using a method corresponding to the compression method, and f For example, predetermined data is weighted and added to all of the overlapping sub-blocks.

In this case, each decompression circuit 41-43 is a compression circuit 3B-3.
8, and compression/expansion processing is performed using the same method, but the compression/expansion method in each band does not necessarily have to be the same. That is,
In the set of compression circuit 36 and expansion circuit 41, DPCM
The compression circuit 38 and expansion circuit 4
It is possible to freely select the compression/expansion method to take advantage of the characteristics of each band, such as using the vector quantization method in the group No. 3.

The image data expanded for each band by the expansion circuits 41 to 43 is then processed by the subblock synthesis circuits 44 to 4B.
The image data is synthesized by the filter circuits 30 to 32 and restored into image data for each band separated by the filter circuits 30 to 32.

Then, the image data for each band output from each sub-block synthesis circuit 44 to 4B is synthesized by a synthesis circuit 47, restored to the original original image data, and taken out from an output terminal 48 for image display. .

By performing the above processing, the overlapping parts of the high-frequency sub-blocks of the original image data will be reduced, which will reduce the amount of transmitted data, and block distortion will become less visually noticeable in the reproduced image, reducing distortion. It is possible to obtain high-quality images with fewer images. In FIG. 5, the original image data is separated into three band components, but it goes without saying that the number of band divisions is not limited to three and may be any number.

Next, FIG. 7 shows a third embodiment of the present invention. First, original image data supplied to the input terminal 49 is supplied to subblock division circuits 50 and 51 and divided into a plurality of subblocks. In this case, the sub-block dividing circuit 50 is for DC component processing, and performs the dividing process so that each divided sub-block has an overlapping portion of a predetermined width around its periphery.

Further, the sub-block dividing circuit 51 is for AC component processing, and performs division processing so that each divided sub-block has no overlapping portion.

The size of each sub-block having no overlapping portions divided by the sub-block dividing circuit 51 is set to be a square of N×N pixels as shown in FIG. 8 in consideration of later processing. Furthermore, each sub-block divided by the sub-block dividing circuit 50 has a width C at the periphery as shown in FIG.
It is assumed that there is an overlapping part. Note that the centers of each subblock divided by both subblock division circuits 50 and 51 are at the same position.

Now, if N-8 is used, an 8x8 pixel sub-block with no overlapping parts will be used for AC, and if it is C-2, the AC sub-block will be one pixel larger on the outside for DC. A sub-block of IOX 10 pixels will be used.

Then, for the data of each sub-block divided by the sub-block dividing circuit 50, an average value of each sub-block is calculated by an average value circuit 52. This average value is the direct current component of each subblock divided by the subblock division circuit 5o, and only one value is calculated for each subblock.

This DC component is compressed by the compression circuit 53, but here, the difference between adjacent subblocks is taken and compression is performed by D.
PCM compression processing is used.

Further, the average value of each sub-block is subtracted by a subtracting circuit 54 from the data divided into sub-blocks without overlapping portions by the sub-block dividing circuit 51. That is, from the subblock data obtained from the subblock division circuit 51, only the average value, that is, the DC component is removed.

However, since the sizes of sub-blocks with overlapping parts and sub-blocks without overlapping parts are slightly different, the average values of the two do not match completely, but the difference is small and there is no difference in practical luck. There isn't.

The original image data from which the average value has been removed by the subtraction circuit 54 is compressed by the compression circuit 55.

Here, the compression method of the compression circuit 55 is a transform method using orthogonal transform such as Fourier transform or cosine transform. The feature of this method is that the spatial information of the image is converted into spatial frequency information. The AC component of the data compressed by the compression circuit 55 is extracted by the AC component extraction circuit 5B.

In this case, the data that has been orthogonally transformed by the compression circuit 55 includes:
Since no DC component is included, the DC component value is zea. Therefore, even if the AC component extraction circuit 56 extracts only the AC component, no problem will occur in the image information. In this way, the DC component compressed data obtained from the compression circuit 53 and the AC component compressed data obtained from the AC component extraction circuit 5B are synchronized by the synchronization circuit 57 and output as one image data. Ru.

The image data compressed in this way is
After being recorded on a predetermined recording medium or transmitted, it is supplied to a separation circuit 58 on the reproduction side. This separation circuit 58 separates the input image data into DC component compressed data and AC component compressed data. Then, the compressed data of the DC component is expanded by the DPCM method by an expansion circuit 59, and then subjected to subblock synthesis by a subblock synthesis circuit 60. In this case, the compressed data of the DC component is
Although the sub-blocks have an overlapping portion, weighting is performed on this overlapping portion by addition in the same manner as in each of the embodiments described above.

Further, the compressed data of the AC component is expanded by an expansion circuit 61 using an orthogonal transformation method. At this time, since the transmitted compressed data does not include a DC component, the DC component is treated as '0'.

The data converted into spatial information by the orthogonal transformation of the decompression circuit 6L is subjected to subblock synthesis by the subblock synthesis circuit 62. In this case, since the AC component compressed data does not have any overlapping parts in the subblocks, no weighting is performed.

In this way, the DC component and AC component that have been independently processed are combined by the combining circuit B3 and restored to the original original image data, which is taken out from the output terminal θ4 and used for image display.

According to the above embodiment 'N43, it is possible to eliminate block distortion of the DC component that is most visually noticeable, and to improve the image quality of the reproduced image. Also, one important point of this third embodiment is that only one DC component exists in each sub-block, and whether the sub-blocks have overlapping parts or not, the total information is This means that the amount remains unchanged.

In other words, according to the third embodiment, image quality can be improved without increasing the amount of information.

It should be noted that the present invention is not limited to the above-described embodiments, and can be implemented with various modifications without departing from the gist thereof.

[Effects of the Invention] As detailed above, according to the present invention, even if an original image is divided into a plurality of blocks and each block is independently compressed and expanded, block distortion does not occur in the reproduced image. It is possible to provide an extremely good image data processing device that can perform the following operations.

[Brief explanation of the drawing]

FIG. 1 shows an embodiment of an image data processing device according to the present invention, and is a diagram for explaining sub-block division.
FIG. 2 is a block configuration diagram showing a configuration for realizing the same embodiment, FIG. 3 is a diagram for explaining weighting in the same embodiment, and FIG. 4 shows a modification of the same embodiment. FIG. 5 is a block configuration diagram showing a second embodiment of the present invention; FIG. 6 is a characteristic diagram showing the characteristics of the filter circuit of the second embodiment; FIG. 7 is a block configuration diagram showing a third embodiment of the present invention, FIGS. 8 and 9 are diagrams for explaining sub-block division of the third embodiment, and FIG. 10 is a conventional image. FIG. 2 is a diagram conceptually showing compression and expansion of data. ...Original image, 12...Sub block, 13...Input terminal, 14...Sub block division circuit, 15-23
...Sub block, 24...Compression circuit, 25...
Decompression circuit, 2B... Sub-block synthesis circuit, 27.
...Output terminal, 28...Circular sub-block, 29...
・Input terminal, 30-32...Filter circuit, 33-3
5... Sub-block division circuit, 36-38... Compression circuit, 39... Simultaneous f-hi circuit, 40... Separation circuit,
41-43...Extension circuit, 44-46...Sub-block synthesis circuit, 47...Composition circuit, 4B...Output terminal, 49...Input terminal, 50.51...Sub-block division Circuit, 52... Average value circuit, 53... Compression circuit, 54... Subtraction circuit, 55... Compression circuit, 5B.
...AC component extraction circuit, 57...Synchronization circuit, 58.
...Separation circuit, 59...Extension circuit, 60...Sub block synthesis circuit, 81...Extension circuit, 62...Sub block synthesis circuit, B3...Synthesis circuit, B-to...Output terminal . Applicant's agent Patent attorney Takehiko Suzue Engraving of the drawings (no changes to the contents) Figure 3 Figure 4 Profit diagram Ink diagram Figure 9 Procedural amendment 1989 4. n4th Japan Patent Office Commissioner Furu 1) Moon Yi 1, Incident Display Patent Application No. 1-47220 4゜Relationship with the case Patent applicant (307) Toshiba Corporation

Claims (3)

[Claims] (1) Divide the original image into multiple blocks, compress the amount of data for each block, expand the compressed data for each block, and combine the expanded blocks to obtain the original image. The image data processing device includes: a dividing means for dividing the original image into a plurality of blocks such that their peripheral parts overlap each other; a compression means for independently compressing the amount of data on each block divided by the dividing means; The output of the compression means is expanded independently for each block, and in the overlapping portion of each block, there is an expansion means that weights and adds the overlapping blocks, and each block expanded by this expansion means is combined and 1. An image data processing device comprising: a composition means for obtaining an original image. (2) The image data according to claim 1, wherein the dividing means increases the amount of overlapping of each block in a low frequency range of the original image, and decreases the amount of overlapping of each block as the frequency range becomes higher. Processing equipment. (3) The dividing means divides the original image into a DC component and an AC component, dividing the DC component into a plurality of blocks so that their peripheral parts overlap each other, and dividing the AC component into a plurality of blocks so that their peripheral parts do not overlap. The image data processing apparatus according to claim 1, wherein the image data processing apparatus is divided into blocks.