JPH0421274A - Picture processing system - Google Patents

Abstract

PURPOSE:To obtain a reproduced picture data with a few bit numbers and less deterioration by decoding a data in response to the statistic property specific to each element information according to information relating to a picture compression system such as subblock processing information or quantity information. CONSTITUTION:An inverse subblock processing setting circuit 14 reads subblock processing information 9 in a code data D to set an inverse subblock processing device 16 and an inverse quantizer setting circuit 1 reads quantizer information 10 in the code data D to set an inverse quantizer 17. The inverse quantizer 17 obtains converts a Y component compression data 11, an I component compression data 12 and Q component compression data 13 in the code data D respectively into differences of the component and adds them to obtain a Y component 21, an I component 22 and a Q component 23. Then a Y component reproduction data 24, an I component reproduction data 25, and a Q component reproduction data 26 are converted into an R component 27, a G component 28 and a B component 29 by a YIQ-RGV conversion circuit 19.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to an image processing method, and more specifically, a signal compression method necessary for transmitting or recording image information, and a compressed digital image. A decoding method that decodes the analog reproduction image signal (2 decoding methods [prior art]) Image information, especially color images, has a huge amount of data, so in order to realize efficient data transmission or data recording, it is necessary to improve the compression ratio. requires a large data compression method and a data decoding method that can reproduce clear images.

Conventionally, as shown in Japanese Patent Application Laid-open No. 64-53680, a conventional data compression method separates and encodes a color image into two or more elemental signals, and then encodes these elemental signals. A method is known in which two or more different compression methods are applied.

Other methods include, for example, Japanese Patent Application Laid-Open No. 63-1589
As shown in Japanese Patent Application No. 71, variable sampling density and DPCM, which are characterized in that a color image is separated into a luminance signal and a color difference signal, and the color difference signal is compressed with coarser pixels than the luminance signal, are used. A method is known that is characterized by compression using a quantization method such as

[Problem to be solved by the invention]

However, in the former of the conventional examples, when reproducing images from compressed data, it is necessary to prepare various decoders suitable for decoding data compressed by various methods, so the reproduction device is complicated and The problem is that it is expensive.

On the other hand, in the latter case, when the sampling interval is changed with a variable sampling density, as the sampling interval becomes longer, the number of pixels that are not subject to compression increases, so the reproduced image is likely to deteriorate.On the other hand, when the sampling interval is shortened, the data There is a problem in that the compression ratio of the data decreases and efficient data transmission and data recording cannot be performed.

The present invention has been made in order to solve the problems of the prior art, and has a simple configuration, a high compression ratio, and a system that sufficiently reflects the statistical properties specific to the element information of the original image. It is an object of the present invention to provide an image processing method suitable for obtaining reproduced images with little deterioration in image quality.

[Means for Solving the Problems] In order to achieve the above object, the present invention provides an image processing method that separates an image into a plurality of element information, compresses each element information, and obtains compressed data. Data is compressed using the same compression method and a sub-blocking method and quantization method suitable for each element information, and information about the sub-blocking method and quantization method for each element information is added to the resulting compressed data. is added to obtain code data.

In addition, multiple pieces of element information separated from an image are compressed using the same compression method and a sub-blocking method and quantization method suitable for each piece of element information, and the resulting compressed data is used for each piece of element information. In an image processing method that decodes a reproduced image from coded data to which information regarding a subblocking method and a quantization method is added, a subblocking method and a quantization method applied when data compressing each element information from the coded data. The compressed data is decoded by reading information related to the compressed data and applying the corresponding desubblocking method and dequantization method.

[Operation] By adding information about the image compression method, such as sub-blocking information or quantization information, to the compressed data, the image can be compressed according to the statistical properties specific to each element information of the original image, and then a simple It is possible to decrypt using this method.

In addition, by performing image compression according to the statistical properties unique to each elemental information of the original image, it is possible to optimize the quantizer to a state unique to each elemental information of the original image, reducing the number of bits. Therefore, it is possible to obtain a reproduced image with less deterioration.

Furthermore, when creating sub-blocks, pixels with element information that does not change much are integrated to create sub-blocks.
It is possible to increase the compression rate while suppressing the deterioration of reproduced images.

〔Example〕

FIG. 1 is a block diagram of an image compression device used to implement the color image processing method according to the present invention,
B-Y IQ conversion circuit 1, difference value generation circuit 2, subblock generator setting circuit 3, quantizer/inverse quantizer setting circuit 4, subblock generator 5, and quantizer 6 , an inverse quantizer 7 , and a delay device 8 .

RGB-Y IQ conversion circuit 1 has R (red), G (green)
, B (blue) components is converted into a luminance component Y and color difference components I and Q.

The difference value generation circuit 2 obtains the difference value between adjacent pixels from the 20 Y, I, and Q components for each component, and uses a subblock generator 5, a quantizer 6, and an inverse quantizer according to the statistical properties. Perform the settings in step 7.

Settings for the subblock generator 5 are performed through the subblockizer setting circuit 3, and settings for the quantizer 6 and the dequantizer 7 are performed through the quantizer/inverse quantizer setting circuit 4.

The sub-block converter 5 converts the Y component, ■ component, and Q component of the original image.
The components are integrated into subblocks in appropriate pixel units set by the subblock generator setting circuit 3.

The quantizer 6 converts the Y component, engineering component, and Q component digital signals integrated into subblocks by the subblock generator 5 into appropriate thresholds set by the quantizer/inverse quantizer setting circuit 4. Make compressed data by value. Specifically, D
A PCM data compressor or the like can be used.

The dequantizer 7 represents the compressed data compressed by the quantizer 6 with an appropriate representative value.

The delay device 8 delays the output signal of the inverse quantizer 7 and sequentially adds it to the output signal of the subblock generator 5.

The principles of the sub-blockizer setting circuit 3 and the quantizer/inverse quantizer setting circuit 4 will be explained below.

Natural images are usually expressed using three primary colors, R, G, and B, but due to human visual characteristics, they can be separated into luminance components and color difference components by matrix conversion.

As a matrix for obtaining the luminance component and the color difference component from the R, G, and B components, for example, the following formula 1 is used.

■.

The Q component has a strong correlation between adjacent pixels, and data compression is possible by removing redundancy due to this correlation. In addition, the color difference components I and Q are
Since the visual component is relatively low, the requirements for reproducibility are not very high. Furthermore, the correlation between adjacent pixels is determined by the color difference component 1.
Q is generally stronger than the luminance component Y, and changes in density between adjacent pixels are small. Therefore, the compression ratio of the color difference components I and Q can be increased compared to the luminance component Y.

In the present invention, in order to perform such data compression, sub-blocking is performed according to the statistical properties of the difference values between adjacent pixels of each element information Y, I, and Q of the original image.

Furthermore, in order to reduce deterioration of the reproduced image, quantization is performed according to the statistical properties of the difference values between adjacent pixels of each element information Y, I, and Q of the original image.

That is, in FIG. 1, the R, G, and B input image data is converted into a luminance component Y and color difference components I and Q by an RGB-YIQ conversion circuit 1, and then a difference value generation circuit 2 converts the difference between adjacent pixels. get value This difference value between adjacent pixels is generally the fourth
Since the distribution shown in the figure is taken, the desired subblock and quantization can be performed by setting the subblock generator 5, quantizer 6, and inverse quantizer 7 according to this distribution. can.

Next, a method of dividing element information into sub-blocks will be explained based on FIG. 5.

FIG. 5 is an explanatory diagram showing an example of sub-block formation, and shows an example of sub-block formation in which a sub-block is created by integrating three pixels in the horizontal direction and two pixels in the vertical direction. Of course, the size of the sub-block is not limited to this, but is determined by the statistical properties of the difference values between adjacent pixels.

The representative density of a subblock can be determined by the average value of the density of each pixel included in the subblock.

Another method is to take advantage of the fact that the densities of each pixel in a sub-block have a strong correlation and that there is a high probability that there are multiple pixels with the same density. Concentrations can also be determined.

Still another method is to use the median value of the densities of each pixel included in a sub-block as the representative density of the sub-block.

In addition, it is also possible to approximate the density of pixels in a sub-block by an appropriate function, and use an appropriate variable as the representative density.

In any case, since the subblocking method of the present invention targets all pixels in a subblock and determines the representative density of the subblock according to a certain rule, as in the case of sampling, for example, the subblock is ignored. Since there are no pixels at all, deterioration of the reproduced image can be reduced.

Next, the meaning of sub-blocking according to the statistical properties of the difference values between adjacent pixels of each element information of the original image will be explained.

Basically, in areas where there is a strong density correlation between adjacent pixels, the deterioration of the reproduced image is not noticeable even if a large sub-block is created by integrating many pixels.

The present invention utilizes this property and attempts to obtain a reproduced image with a high compression rate and little deterioration by appropriately changing the size of a subblock according to the correlation of density between adjacent pixels.

As an index indicating the correlation of density between adjacent pixels, for example, the standard deviation of difference values between adjacent pixels can be used. In this case, the smaller the standard deviation of the difference values, the larger the sub-blocks can be made.

Furthermore, as another method, the probability that the difference value between adjacent pixels is O (zero) may be used as an index indicating the correlation of density between adjacent pixels. In this case, the larger the probability of O, the larger the sub-block may be.

Furthermore, the standard deviation of the difference value between adjacent pixels and the probability that the difference value between adjacent pixels is ◯ may both be considered as an index indicating the correlation of density between adjacent pixels.

Next, a quantization method according to the statistical properties of the difference value between adjacent pixels of each element information of the original image will be explained.

As shown in FIG. 4, the frequency of the difference values increases as the difference values approach ◯ for all of the Y component, ■ component, and Q component. Therefore, by quantizing finely as the difference value approaches O, data compression can be performed using the statistical properties of the difference value.

Now, as shown in FIG. 4, the probability distribution of the difference values of the luminance component Y and the probability distribution of the difference values of the color difference components 1 and Q are different.

The probability distributions corresponding to each component of Y, I, and Q are Pv, P+, P
When Q is known as M a x quantization, create a nonlinear quantizer that minimizes the mean square error 3' between the input and output of the quantizer expressed by the second equation below. It is possible to do so.

However, P (z) = PY (z) Y component = pl
(z) ■Component = PQ (Z) K in the second equation of Q component is the number of output levels, qI.

qK is the output level, Zll ZII ...Z,
+1 is the judgment level, and P (z) is the probability distribution for the input sample value represented by the continuous variable Z.

For a given number of output levels K, the mean square error ε“
The quantizer can be optimized according to the probability distribution by setting the output levels q, q, . . . ql and the decision levels ZllZll .

In the present invention, as the quantizer 6, for example, the above-mentioned M a
Using an x quantizer, this is set based on the probability distribution of difference values between adjacent pixels, and an inverse quantizer 7 generates compressed data that can be inversely quantized.

Note that in order to reduce the calculation load when setting the M a x quantizer, it is preferable to share the number of output levels for quantization of each element information. In addition, in order to reduce the calculation burden, an average quantizer is created in advance based on several sample images, and the statistical properties of the difference values between adjacent pixels of the target image are calculated.

For example, it is also possible to correct the quantizer using standard deviation or the like as a guide.

Image compression processing using the image compression apparatus shown in FIG. 1 will be described below.

The R, G, and B components of a color image are converted into a luminance component Y and color difference components I and Q by an RGB-Y IQ conversion circuit 1.

A difference value between adjacent pixels is obtained for each component from the second Y, I, and Q components by a difference value generation circuit 2, and a subblock generator 5, a quantizer 6, and an inverse quantizer 7 are used according to the statistical properties. Configure settings.

The sub-blocking device setting circuit 3 is used to set the sub-blocking device 5, and information necessary for reverse sub-blocking is made part of the code data D as sub-blocking information 9. Similarly, the settings for quantizer 6 and inverse quantizer 7 include quantizer/
Using the dequantizer setting circuit 4, information necessary for dequantization is made part of the code data D as quantization information 10.

After that, the Y component output from the subblock generator 5,
■ Compress and encode each of the components and Q components using a predetermined data compression method, such as the DPCM (differential PCM) method.
Component compressed data 11, ① component compressed data 12, and Q component compressed data 13 are obtained, and these are made into part of code data D.

Next, a compressed data decoding method according to the present invention will be explained.

FIG. 2 is a block diagram of an image reproducing device that decodes compressed data to obtain a reproduced image, and includes a desubblock generator setting circuit 14, a dequantizer setting circuit 15, and a desubblock generator 16. , the inverse quantizer 17, the delay device 18, and Y
It mainly consists of an IQ-RGB conversion circuit 19 and a correction filter 20.

The inverse sub-blocking device setting circuit 14 reads the sub-blocking information 9 in the encoded data D and converts the sub-blocks created by the sub-blocking device 5 in FIG. 1 into pixel units. Set up the blocking device 16.

The inverse quantizer setting circuit 15 reads the quantization information 10 in the encoded data D, and extracts the Y component 21 and the ■ component 2 from the compressed data 11, 12, and 13 compressed by the quantizer 6 in FIG.
2. Set the inverse quantizer 17 so that the Q component 23 can be decoded.

The inverse sub-block generator 16 is one that obtains the densities of all pixels in the sub-block from the representative densities of the sub-blocks, corresponding to the sub-blocks set by the sub-block generator 5 of FIG.

For example, when the sub-block generator 5 is one that determines the representative density of a sub-block from the average density of all pixels in the sub-block, the inverse sub-block generator 1
6, the density of all pixels in the sub-block is used as the representative density of the reproduced sub-block. Also,
When the subblock generator 5 is used, which approximates the pixel density in the subblock by an appropriate function and uses an appropriate variable as the representative density when forming the subblock, the inverse subblock is created by the corresponding inverse transformation. A reverse sub-block generator 16 that can perform blocking is set up. The same applies to the case where the sub-block generator 5 which determines the representative density by another method is used.

The inverse quantizer 17 used has the same characteristics as the inverse quantizer 7 set in the image compression apparatus shown in FIG.

The Y I Q-RGB conversion circuit 19 converts the Y component signal 24 and ■ component signal 25 output from the inverse subblock generator 16 into
, the Q component signal 26 is converted into an R component 27, a G component 28, and a B component 29.

Conversion by the Y IQ-RGB conversion circuit 19 is performed according to the third equation.

(3) The correction filter 20 converts the R component 27, G component 28, and B component 29 obtained by the YIQ-RGB conversion circuit 19 so that the values of each component are within the allowable density range. Correct to. This correction filter 20 corrects the overflowed numerical value.

Image processing using the image reproducing apparatus shown in FIG. 2 will be described below.

First, the desubblock generator setting circuit 14 sets the code data D
At the same time, the dequantizer setting circuit 1 reads the quantization information lO in the encoded data D and sets the dequantizer 17. .

The inverse quantizer 17 returns the Y component compressed data 1U, the ■component compressed data 12, and the Q component compressed data 13 in the encoded data D to component difference values, and adds these to the Y component. Data 21, ■ component data 22, and Q component data 23 are obtained.

Next, the Y component data 21, the ■ component data 22, and the Q component data 23 are converted into reverse subblocks by the reverse subblock generator 16, and the Y component reproduced data 24 and the ■component reproduced data 25 are
, obtain Q component reproduction data 26.

Thereafter, the Y component reproduced data 24, the ■ component reproduced data 25, and the Q component reproduced data 26 are converted into an R component 27, a G component 28, and a B component 29 in a YIQ-RGB conversion circuit 19.

Finally, the R component 27, G component 28, and B component 29 are applied to the correction filter 20, and the target R component reproduced image data 3
C), G component reproduced image data 31 and B component reproduced image data 32 are obtained.

In the above embodiment, D is used as the data compression method.
Although the PCM method was used, other known data compression methods may also be used.

Furthermore, in the embodiment, according to the first equation, R,
Convert the G and B components to Y, I and Q components, and convert the Y, I and Q components to R and G according to the third equation.

Although the B component is converted, conversion according to other conversion methods can also be performed.

Furthermore, in the embodiment, R, G,
The B component is divided into the luminance component Y and the color difference component 1. I converted it to Q, but
It is also possible to convert to element information other than Y, I, and Q.

An example of an image processing system using the above-described image compression device and image reproduction device will be described below with reference to FIG.

As is clear from this figure, the input side of this system includes a color video camera 33, a video signal input circuit 34,
Input image memory 35 and image compression device 3 shown in FIG.
9, and the reproduction side is composed of an image reproduction device 41 shown in FIG. 2, an output image memory 45, a video signal output circuit 46, and a color image display section 47. The code data D exchanged between the input side and the playback side of this system is, of course, the code data D shown in FIGS. 1 and 2.
is the same as

A color image captured by the color video camera 33 is separated into an R component digital signal 36, a G component digital signal 37, and a B component digital signal 38 by a video signal input circuit 34, and stored in an input image memory 35.

The R, G, and B digital signals 36, 37, and 38 stored in the input image memory 35 are processed by the image compression device 3 of the present invention.
9, it is converted into code data D. This code data D is a digital signal that can be stored in an information recording medium such as an optical disc.

On the playback side, the image playback device 4 of the present invention uses the encoded data D.
1, R component reproduced image data 30, G component reproduced image data 31, and B component reproduced image data 32 are obtained, and the reproduced images are displayed on a color image display section 47 via an output image memory 45 and a video signal output circuit 46.

[Effects of the Invention J As explained above, according to the present invention, data compression is performed according to the statistical properties unique to each piece of elemental information constituting the original image, and sub-data is added to the compressed data obtained thereby. Information about the image compression method, such as blocking information or quantization information, is added, and the playback side uses this information to decode data according to the statistical properties unique to each element information, so the number of bits is small. It is possible to obtain reproduced images with less deterioration.

In addition, since each element information of the original image is compressed using the same compression method, decoding of each element information can basically be performed in common with one decoding device, which simplifies the device and reduces playback time. It is possible to shorten the time.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of an image compression device according to the present invention, FIG. 2 is a block diagram showing an example of an image reproducing device according to the present invention, and FIG. 3 is a block diagram showing an example of an image compression device and an image reproducing device according to the present invention. FIG. 4 is a graph diagram showing the distribution of difference values between adjacent images regarding each element information of the original image, and FIG. 5 is an explanatory diagram of sub-blocking. 1...RGB-YIQ conversion circuit, 2...
- Difference value generation circuit, 3... Sub-block generator setting circuit, 4... Quantizer/inverse quantizer setting circuit, 5... Sub-block generator, 6... ... Quantizer, 7 ... Inverse quantizer, 8 ... Delay device, 9 ... Subblock information, 1o ...
・Quantization information, 11...Y component compressed data, 1
2... I component compressed data, 13... Q component compressed data, 14... Inverse subblock generator setting circuit, 15... Inverse quantizer setting circuit, 16・
...Inverse sub-block generator, 17...Inverse quantizer, 18...Delay unit, 19...Y I
Q-RGB conversion circuit, 20...correction filter,
D... Code data. Fig. Fig. Fig. Fig. Fig. Diagram J

Claims (3)

[Claims] (1) In an image processing method that separates an image into multiple pieces of element information and compresses each piece of element information to obtain compressed data, each piece of element information is compressed using the same compression method, and sub-blocks suitable for each piece of element information are used. An image processing method characterized by compressing data using a quantization method and a quantization method, and adding information regarding a subblocking method and a quantization method for each element information to the obtained compressed data to obtain encoded data. (2) Data compress multiple pieces of element information separated from an image using the same compression method and a sub-blocking method and quantization method suitable for each element information, and add each element information to the resulting compressed data. In an image processing method that decodes a reproduced image from coded data to which information regarding a subblocking method and a quantization method is added, the subblocking method and quantization applied when compressing each element information from the coded data. An image processing method characterized in that compressed data is decoded by reading information regarding the method and applying a corresponding desubblocking method and dequantization method. (3) In claim 1 or 2, the image is a color image, and the luminance component and two color difference components of the color image are data compressed using a DPCM method, and the data is compressed according to the distribution of adjacent difference values of each element information. subblock generator, quantizer,
An image processing method characterized by adjusting a desubblock generator and a dequantizer.