JPH11164320A - Method and system for decoding color compression image - Google Patents

Abstract

PROBLEM TO BE SOLVED: To decode a visual image that has a requirement of high image quality with respect to saturation like a photo print from a compressed color image. SOLUTION: In the case of decoding a compressed color image, the saturation is corrected by a correction amount depending on a compression rate of a color component to compensate color information lost in the case of compression. For example, the correction amount is decided depending on whether or not pixel number reduction processing is applied to a color difference component in the compression processing and which quantization table is used (represented by a Quality index) in the case of quantization.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for decompressing (expanding) a color image compressed by, for example, JPEG or the like and reproducing a color compressed image reproduced as a photographic print.

[0002]

2. Description of the Related Art Conventionally, orthogonal transformation, quantization, entropy coding and the like have been known as image compression techniques. The orthogonal transform focuses on the fact that image information having a strong correlation is concentrated in the low frequency region when viewed on the frequency axis, and only the frequency components in which the information is concentrated and distributed are considered. This is a compression technique that reduces the amount of information as a whole by encoding. Quantization is a compression technique that divides image information by a certain numerical value (quantization step) and reduces the number of bits representing the image information by approximation at a limited number of discrete levels. In addition, entropy coding is a compression technique that reduces the total code amount by assigning a short code to a value with a high occurrence frequency and a long code to a value with a low occurrence frequency when assigning a binary code to image information. is there.

[0003] JPEG, which is widely known as a coding standard for color still images, encodes image information by combining the above-mentioned compression techniques. The image block is divided into image blocks of pixels, and a discrete cosine transform (DCT), which is one of orthogonal transforms, is performed on each image block. A transform coefficient obtained by the DCT is quantized. Image information is compressed by performing Huffman coding, which is entropy coding, on the values.

[0004] An image compressed by JPEG can be decompressed by performing the reverse procedure of the above procedure, ie, entropy decoding, inverse quantization, and inverse discrete cosine transform (IDCT).

Further, since human vision is sensitive to luminance and insensitive to color difference, image information is separated into a luminance component and a chrominance component by performing color conversion from RGB space to YCC space. A method of individually performing the above-described compression processing for each component at a low compression ratio for the components and a high compression ratio for the color difference components has also been proposed.

[0006]

In recent years, the digitization and networking of photo services have been advanced, and from the above-mentioned compressed images, high-quality images of the same level as conventional photo prints (prints made from film) have been obtained. There has been a demand for making photo prints.

[0007] However, the above-mentioned image compression techniques include:
Because it is a technology developed for the purpose of efficiently transmitting or storing images handled on a computer, the image quality when decompressing a compressed image should be sufficient if it is sufficient image quality on a monitor screen, and it is not necessarily The image quality is not enough for making photo prints.

In particular, a photographic image, unlike an image generated by computer graphics, does not have a vivid color portion filled with a single vivid color. In an area that is visually perceived as one vivid color in a photographic image, various colors that are close to the vivid color exist in a microscopic manner, and the human eye recognizes those colors. The sum of the colors appears as one bright color.

However, in the above-described compression processing, since the compression ratio is increased by reducing high-frequency information that carries the complexity of the image, the colors of the image are averaged as the compression ratio is increased. For example, even if a region of a predetermined color in a pre-compressed photographic image includes a pixel having a particularly vivid color, once the photographic image is compressed and decompressed,
The bright colors are lost and averaged to the same color as the surroundings. The higher the compression ratio, the more color complexity is lost. As a result, a photographic print created by restoring a compressed image is slightly faded as compared with a case where an image before compression is printed as it is.

SUMMARY OF THE INVENTION In view of the above problems, the present invention compensates for color information lost during compression at the time of decompression of a compressed image, thereby preventing a decrease in saturation caused by repeated compression and decompression. It is an object of the present invention to realize a high image quality that can withstand the creation of a photographic print for the obtained image.

[0011]

According to the present invention, in order to solve the above-mentioned problems, a correction for increasing the saturation at the time of decompression of a compressed image is performed. Since saturation differs depending on the compression method, saturation is corrected with a correction amount appropriate for the compression method.

That is, a method of restoring a color-compressed image according to the present invention is a method of restoring a color-compressed image in which a color image compressed by a predetermined compression method is expanded and reproduced as a visible image. And performing a correction process to increase the saturation of the color image with a correction amount depending on the compression ratio of the color difference component of the color image by the compression method, and the corrected color image is recorded on a recording medium. Alternatively, the image is reproduced as a visible image on a display medium.

Here, in the expansion processing and the correction processing, the correction processing may be performed after the expansion processing is performed, or the correction may be performed in the process of the expansion processing. For example, an image compressed by JPEG is decompressed stepwise by performing a plurality of decompression processes such as Huffman inverse coding, inverse quantization, and inverse DCT. In this case, a process of increasing the saturation may be performed at an intermediate stage, for example, when the inverse quantization is completed.

The "compression rate of the color difference component" which is a reference when determining the correction amount is determined by, for example, performing a color conversion to decompose the image information into a luminance component and a color difference component and then individually compressing them. What is necessary is just to set the compression ratio of the compression processing performed on the separated color difference components. Specifically, for example, in the process of compressing the color difference components by quantization, the correction amount may be determined for each type of quantization table for quantization. In other words, since the compression ratio increases as the quantization step increases, the amount of correction added to increase the saturation of an image compressed using a quantization table with a large quantization step is increased. is there.

The compression ratio of the color difference component cannot always be quantitatively defined in the case of a compression method in which the color difference component is not separated during the compression process. In such a case, the compression method is roughly classified, for example, a compression method in which the saturation is remarkably reduced, a compression method in which the saturation does not change much, and the correction amount is determined based on the classification. Good.

It is desirable that the "correction amount" be determined in advance for each compression method. For example, at the time of compression, the name of the compression method or other parameters for specifying the compression method are described in a file header or the like of the image file. On the other hand, a table in which the compression method and the correction amount are associated with each other is stored in the memory of the device that decompresses the image file. Thus, when the image is decompressed, the method of compressing the image is checked by referring to the information in the file header, and the corresponding correction amount can be read from the memory to perform the correction processing.

Alternatively, the correction amount may be determined at the time of thawing without being stored in advance. For example, in the method of determining the correction amount for each type of quantization table, a method of determining the correction amount by performing a predetermined operation based on the value of a parameter that specifies the type of quantization table can be considered.

Note that the recording medium is photographic paper or the like used for making a photographic print. The display medium is a monitor screen or the like. In monitor display images, generally, the decrease in saturation due to compression is not often a problem. However, when high image quality is required, such as when a part of a photograph is enlarged and displayed on a monitor with a high resolution, the present invention is applied. Is effective.

The apparatus for restoring a color compressed image of the present invention is an apparatus for restoring an image based on the above method. That is, this apparatus is a color compression image restoration apparatus for expanding a color image compressed by a predetermined compression method and reproducing it as a visible image, and expands the color image by an expansion method corresponding to the compression method. Image decompression means, saturation correction means for performing correction processing for increasing the saturation of the color image with a correction amount depending on the compression ratio of the color difference component of the color image by the compression method, and a storage medium for storing the corrected color image on a recording medium Alternatively, an image reproducing means for reproducing a visible image on a display medium is provided.

Here, "saturation correction means" means, for example, when the correction amount is stored in advance, means for reading out the necessary correction amount from the memory and adding it to the image data. . Further, when the correction amount is obtained by calculation at the time of image decompression, it means such a calculating means and a means for adding the correction amount obtained by the calculating means to the image data.

At this time, the correction amount stored by the saturation correction means or obtained by calculation is, for example, after the compression method separates the color difference component of the color image from the color image.
In the case of performing a compression process on the separated color difference components, the method may be determined depending on the compression ratio of the color difference components by the compression process. Specifically, for example, in the process of compressing the color difference components by quantizing, the correction amount may be determined for each type of quantization table of quantization.

[0022]

The method and apparatus for restoring a color compressed image according to the present invention perform a correction such that the saturation is increased with a correction amount suitable for the compression method when the compressed image is decompressed. As a result, color information lost during compression is compensated for when decompressing a compressed image, so that it is possible to prevent a decrease in saturation caused by repeated compression and decompression, and also to create a photographic print. A tolerable high-quality restored image can be obtained.

At this time, since the compression ratio of the chrominance component of the image compressed by the method of separating and compressing the chrominance component is clear, if the saturation correction is performed based on the compression ratio, accurate correction can be performed. It can be carried out. Further, when the color difference components are separated and compressed, the color difference components are handled separately even in the decompression process, so that the color difference components can be directly subjected to the saturation correction. That is, the color difference component can be directly corrected without performing special processing such as color conversion, so that the processing efficiency is high.

In the compression by quantization, since the compression ratio of the image is finely set by a parameter for controlling the quantization table, if the saturation correction amount is determined in accordance with the quantization table, the compression ratio can be reduced. , It is possible to perform a correction that is more appropriate for the reduced saturation.

[0025]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. The embodiment described below relates to restoration of a color image compressed by JPEG.

FIG. 1 is a diagram schematically showing JPEG compression and decompression processing. As shown in the figure, in this process, first, image information is separated into a luminance (Y) component and a color difference (C) component by performing color conversion from an RGB space to a YCC space, and each component image is, for example, 8 ×. The image block is divided into 8-pixel image blocks (not shown), and each image block is subjected to a discrete cosine transform (DCT), which is one of orthogonal transforms, and a transform coefficient obtained by the DCT is quantized. The image information is compressed by performing Huffman coding, which is entropy coding, on the quantized value.

The compressed image is restored by performing the reverse procedure, that is, Huffman inverse coding, inverse quantization, inverse discrete cosine transform (IDCT), and color conversion from YCC space to RGB space.

In this embodiment, the saturation of an image is corrected in accordance with a quantization table used for quantization in the above processing. Therefore, first, the quantization table will be briefly described. In the following description, each piece of information is treated as 8-bit data.
That is, for example, the color difference is usually represented by a value in the range of -1 to 1, but in the following embodiment, the value in the range of -1 to 1 is mapped to an integer value in the range of 0 to 255. It will be described as.

FIG. 2 is a diagram showing the meaning of data (coefficients) obtained as a result of DCT. That is, when DCT is applied to the 8 × 8 image block of each component image described above, the data in the image block is classified according to the spatial frequency band, and as shown in FIG. The lower right part is converted into 8 × 8 coefficients representing a component having a higher spatial frequency.

In the JPEG quantization, the 8 × 8 data is subjected to linear quantization as shown in FIG. 3. In this case, the quantization step is not necessarily performed in all steps shown in FIG. Are not necessarily the same. For this reason, a quantization table showing a list of quantization steps when quantizing each coefficient as shown in FIGS. 4 and 5 is usually defined.

For example, FIG. 4 shows that the quantization step is 1 for all coefficients. In this case, even if quantization is performed, the information amount does not decrease. On the other hand, FIG.
The case where the quantization step is increased as the frequency becomes higher is shown. This is defined based on the premise that high-frequency components do not contain very important information and that the compression ratio may be increased. As described above, the degree of compression by quantization depends on the quantization table.

Here, in the quantization table, it is possible to define each step value individually.
Usually, it is often controlled by a parameter called Quality. For example, the quantization table of FIG. 4 shows a table when Quality = 100, and the quantization table of FIG. 5 shows a table when Quality = 90.

The quantization table has been described above. Hereinafter, a method of the present invention for correcting the saturation depending on the quantization table will be described.

As described above, JPEG compression takes advantage of the fact that human vision is sensitive to luminance and insensitive to color differences.
It is possible to compress only the chrominance component without particularly compressing the luminance component. The embodiment described below relates to a case where an image compressed by such a method is restored.

FIG. 6 is a diagram showing one process of a process for particularly compressing only the color difference component. As shown in the figure, in this method, image information is decomposed into a luminance component (Y component) and a color difference component (C component) by YCC conversion.
After performing T, quantization is performed using a quantization table with a higher Quality. That is, the Y component does not compress much. On the other hand, for the C component, DCT is performed after reducing the number of pixels to ４ by taking an average value of 2 × 2 blocks. Further, in quantization, a quantization table having a low quality is used, and the compression ratio is increased. That is, in this compression method, the overall compression ratio is determined by the Quality value of the quantization table for the Y component and the C component, and whether or not to reduce the number of pixels for the C component. After the quantization, Huffman coding is performed as in FIG. 1 (not shown).

FIG. 7 is a diagram showing one process of a process of restoring an image compressed by the method of FIG. Each component of the image information is Huffman inverse-coded as in FIG. 1 and inverse-quantized for each component by a quantization table used at the time of compression (not shown). The inverse DCT is then performed on the Y component. On the other hand, for the C component, when the number of pixels is reduced, interpolation is performed so that the number of pixels is quadrupled, and saturation correction is performed on the interpolated data. Thereafter, the image information is converted to RGB space, and the image is restored.

In the present embodiment, the amount of chroma correction is determined based on whether or not the number of pixels has been reduced during compression and the quality of the quantization table used for quantization of the C component. The parameters of the compression processing, such as whether or not the number of pixels has been reduced and the quality value, are often recorded as supplementary information in the image file of the normally compressed image. On the other hand, the relationship with the correction amount is stored in advance in a memory, for example, in a table as shown in FIG. That is, by referring to this table and obtaining the correction amount corresponding to the compression processing parameter described in the image file to be decompressed, it is possible to perform the saturation correction corresponding to the compression ratio of the color difference component.

As described above, since the value of C (from -1 to 1) is mapped to a value from 0 to 255, if the correction amount described in the table of FIG. Equation (1) below: Cafter = (Cbefore−128) × (1 + H) +128 (1) where Cafter can be performed by the component value after correction, and Cbefore can be performed by the component value before correction.

FIG. 9 shows the correction of the above equation (1). The numerical value in parentheses on the coordinate axis indicates the C
Are shown. That is, since the saturation is higher as the absolute value of C is larger, the saturation can be increased by the equation (1). However, depending on the operation, Cafter <0 or Cafter
If> 255, it is clipped to 0 or 255, respectively, as shown in FIG.

By restoring an image by the above-described restoration method, it is possible to obtain a restored image which does not decrease in saturation as in the conventional method even if compression / decompression is repeated, and which can sufficiently withstand creation of a high quality print. Can be.

Although the restoration of a JPEG-compressed image has been described as an example, it is needless to say that the present invention is not limited to the above embodiment, and can be applied to an image compressed by another compression method. is there.

Also, the saturation correction does not have to be performed directly on the C component.
After conversion into a uniform color space such as LAB (L * a * b) or CIE LUV (L * u * v *), a, b or u, v may be corrected. However, as described above, when the C component is already separated and handled in the decompression processing, the processing efficiency is better if the direct correction is performed because the color conversion processing and the like are not performed.

When creating a photo print,
After the conversion into the RGB space, the color conversion may be further performed on the RGB space suitable for the print output. A plurality of lookup tables used for this conversion are prepared for each saturation correction amount, and each of the compression parameters is prepared. You may switch. That is, it is not always necessary to separate color difference components in order to perform saturation correction.

[Brief description of the drawings]

FIG. 1 is a diagram showing an outline of compression and decompression processing by JPEG.

FIG. 2 is a diagram showing the meaning of data (coefficients) obtained as a result of DCT.

FIG. 3 is a diagram for explaining a linear quantization process;

FIG. 4 is a diagram showing a quantization table of Quality = 100.

FIG. 5 is a diagram showing a quantization table of Quality = 90.

FIG. 6 is a diagram showing one process of a process for particularly compressing only a color difference component.

FIG. 7 is a diagram showing one process of a process of restoring an image compressed by a method of specifically compressing only the color difference component.

FIG. 8 is a diagram illustrating an example of a correction amount table;

FIG. 9 illustrates an example of a correction process.

Continued on the front page (51) Int.Cl. ⁶ Identification code FI H04N 9/808

Claims (6)

[Claims] 1. A method of restoring a color-compressed image, in which a color image compressed by a predetermined compression method is expanded and reproduced as a visible image, wherein the color image is expanded by an expansion method corresponding to the compression method. Performing a correction process to increase the saturation of the color image with a correction amount depending on a compression ratio of a color difference component of the color image by the compression method, and reproducing the corrected color image as a visible image on a recording medium or a display medium. A method for restoring a color compressed image, characterized in that: 2. The method according to claim 1, wherein the compression method separates a color difference component of the color image from the color image and then performs a compression process on the separated color difference component. 2. The method according to claim 1, wherein the amount is determined depending on the compression ratio of the component. 3. The compression processing is processing involving quantization of the color difference component, and the correction amount is an amount determined for each type of quantization table of the quantization. Restoration method of the described color compressed image. 4. A color-compressed image restoration apparatus for expanding a color image compressed by a predetermined compression method and reproducing the color image as a visible image, wherein the color image is expanded by an expansion method corresponding to the compression method. Decompression means; saturation correction means for performing correction processing to increase the saturation of the color image with a correction amount depending on the compression ratio of the color difference component of the color image by the compression method; and An image reproducing means for reproducing a visible image on a display medium; 5. The compression method according to claim 1, wherein the color difference component of the color image is separated from the color image, and then the separated color difference component is subjected to a compression process. 5. The apparatus according to claim 4, wherein the amount is determined depending on a compression ratio of the component. 6. The method according to claim 5, wherein the compression processing is processing involving quantization of the color difference component, and the correction amount is an amount determined for each type of quantization table of the quantization. An apparatus for restoring a color compressed image according to the above.