JPH05205044A - Image color compressing method - Google Patents

Abstract

PURPOSE:To reduce the labor required for compression of an image color and at the same time to shorten the image color compression by automating the visual decision of the picture quality that is hitherto carried out in an image color compressing process. CONSTITUTION:An original image is quantized and a color emerging frequency distribution is decided in a three-dimensional color space. Then plural key color data are decided with preference given to the colors of higher emerging frequency and the color data on each picture element of the original image are replaced with the key color data. Then the color compressed image data are divided into blocks and compared with the original image data for each block. Thus the degree of deterioration of the picture quality is decided, and the picture quality is decided based on the decided degree of deterioration. Then each block is weighted in accordance with the degree of deterioration of the picture quality. These weighted blocks are added to the color emerging frequency distribution and a key color is selected again.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image color compression method for compressing image data of a digital color image, and more particularly to a continuous tone color image such as a natural image in a display device having a color lookup table. The present invention relates to an image color compression method suitable for approximately displaying.

[0002]

2. Description of the Related Art As a conventional image color compression method, there is a method disclosed in Japanese Patent Laid-Open No. 61-129693. In the technology described in this prior application, RGB is used.
The color data appearance frequency distribution of each pixel of the original image in the three-dimensional color space is obtained, the color data with a high appearance frequency is extracted to determine a specific number of representative color data, and the color data of each pixel of the original image is calculated. In the image color compression method for creating image code data obtained by replacing the representative color number of the representative color data,
A color compressed image composed of the representative color data and the image code data is created, and as shown in FIG. 5, a portion having a deterioration in image quality or a point of interest is visually evaluated, and the area is designated by a cursor 501 and weighted. Then, the original image is color-compressed again.

[0003]

However, when the image quality deterioration is visually evaluated as in the above-mentioned prior art, it takes a great deal of effort to color-compress a large number of images, and it is considerably difficult. It will take time.

The present invention has been made in view of the above points,
The purpose thereof is to reduce the labor required for image color compression by automating the image quality evaluation which has been conventionally performed visually, and at the same time to shorten the time required for image color compression.

[0005]

The above-mentioned object is to divide color-compressed image data into blocks, obtain the degree of image quality deterioration for each block by comparing with the original image, and determine the degree of image deterioration in each block according to the degree of deterioration. This is achieved by weighting and color-compressing the original image data again.

[0006]

The entire surface of the color-compressed image is divided into blocks, and the degree of deterioration of the image quality is obtained for each block to determine the position (block) in which the color-compressed image is deteriorated. The color is compressed with a large weight. Then, for example, the image quality evaluation and weighting of each block are repeatedly performed until the desired image quality deterioration is converged, whereby the image quality of the deteriorated portion can be corrected.

[0007]

Embodiments of the present invention will be described below with reference to FIGS. FIG. 1 is a flowchart illustrating an image color compression method according to an exemplary embodiment of the present invention. First, each step of FIG. 1 will be briefly described. Step ST101: a step of quantizing the color data of each pixel of the original image, after which the process proceeds to step ST102. Step ST102: a step of obtaining the appearance frequency of color data, after which the process proceeds to step ST103. Step ST103: a step of determining a representative color,
After this, the process proceeds to step ST104. Step ST104: a step of replacing the color data of each pixel of the original image with a representative color, and then step ST105
Go to. Step ST105: a step of obtaining the image quality deterioration of the color compressed image data for each block, and then step ST1
Proceed to 06. Step ST106: This is a step of performing image quality determination.
If the result of the determination is that the desired image quality deterioration has been reached, this series of processing is terminated, and if not, the operation proceeds to step ST107. Step ST107: a step of weighting the appearance frequency distribution of the color data for each block, after which the process returns to step ST103.

FIG. 2 is a diagram showing an example of an original image and a color compressed image when compressed by the image color compression method in steps ST101 to 104. This figure is an example of an image consisting of mountains, coasts, and yachts. (A) of FIG. 2 shows a pattern of an original image of 8 bits for each of R, G, B, and (b) of FIG.
6 shows a color-compressed image of 6 colors.

The details of the color compression method will be described below for each step. First, in step ST101, the color data of each pixel of the original image is converted into R, G, B 5 bits (32 bits).
Quantization). This is a later step ST103.
The purpose is to avoid selecting only similar colors and to reduce the memory space required for calculation by reducing the color space.

In the next step ST102, the number of appearances of the quantized color data of the original image is obtained, and a color appearance frequency distribution in a three-dimensional color space is created with R, G, and B 5 bits each.
Then, in step ST103, 256 representative colors are selected from the colors having a high appearance frequency in the color appearance frequency distribution of the three-dimensional color space. After that, in step ST104, the color data of each pixel of the original image is calculated in step ST103.
Replace with the representative color number of the closest representative color among the 256 colors selected in.

In this way, a color compressed image in which the original image is replaced with 256 representative colors is obtained. This color-compressed image causes deterioration in image quality when it is replaced with the closest representative color in step ST104. In the example of FIG.
The false contour 201 due to image quality deterioration appears in a part of the sea, and the present invention aims to eliminate such image quality deterioration.

FIG. 3 is a diagram for explaining a method of evaluating the image quality of the color-compressed image created in steps ST101 to ST104 and removing a portion where the image quality is significantly deteriorated. Hereinafter, a method for removing the image quality deterioration (false contour 201) appearing in the sea part will be described with reference to FIG.

In step ST105, the color-compressed image is divided into a total of 24 blocks, for example, 6 horizontal blocks × 4 vertical blocks. Then, the degree of deterioration is calculated for each block in comparison with the original image. Reference numeral 301 in FIG. 3 is one of the blocks, in which the image quality is deteriorated in the sea part, and a false contour which is not present in the original image appears. Examples of the parameter for determining the degree of such image quality deterioration include a standard deviation and a discontinuity evaluation value, but other parameters may be used or these may be used in combination as appropriate.

For example, the standard deviation is obtained by the following equation 1. However, in equation 1, D ......... standard deviation Δr ... difference in luminance from the original image of red component Δg ... difference in luminance from the original image of green component Δb ... of luminance from the original image of blue component Difference n ... The number of pixels in one block.

[0015]

[Equation 1]

When the discontinuity evaluation value is used, the value is obtained according to the following equation 2 (Σ is calculated only when the difference in luminance between adjacent images in the original image is 0 or ± 1). However,
In Equation 2, E ...... Discontinuity value dr ... Difference in luminance of red component of adjacent pixel in compressed image dg ... Difference in luminance of green component of adjacent pixel in compressed image db ... Compressed image The difference in the brightness of the blue component between the adjacent pixels in m ... The number of corresponding pixels (pixels to be calculated) in one block.

[0017]

[Equation 2]

Thereafter, in step ST106, the degree of deterioration of the image quality of each block is determined. In this step, for example, a threshold value for image quality determination is determined in advance, it is determined whether or not the parameters are within the threshold value for all blocks, and if all the blocks are satisfied, the process is terminated and even one If there are blocks that are not satisfied, the process continues. For example, in FIG. 3, the standard deviation D of each block when the image quality deterioration determination threshold value is set to 8.0 by using the standard deviation in the process of step ST105.
The value of is shown as an example for each block.
In the figure, only the block 301 has this value (8.
0) is exceeded, the process continues.

Then, in the next step ST107,
The quantized original image obtained in step ST101 is weighted according to the degree of image quality deterioration of each block (depending on the size of each standard deviation D), and at the same time, step ST1
The color appearance frequency distribution obtained in 02 is added. After that,
Returning to step ST103, the representative color is selected again based on the corrected color appearance frequency distribution, and thereafter, the same processing as described above is executed.

In the following, the color compression processing with the evaluation of the image quality for each block and the weighting according to the evaluation (steps ST103-STEP).
Step ST107) is repeated until the parameters are determined to be within the threshold values for all blocks in step ST106. However, the number of repetitions is limited to a predetermined number because it may occur that the image quality does not converge to a desired image quality. You may do it.

In this embodiment, all blocks are weighted for each block according to the standard deviation D. However, blocks of relatively high image quality whose image quality deterioration determination threshold value is 8.0 or less are originally set. Also, it is not always necessary to weight. Therefore, it is more effective to perform weighting only on the blocks exceeding the threshold value 8.0 according to the value obtained by subtracting the threshold value from the standard deviation D of the block. In this case, if the blocks whose image quality is deteriorated are weighted and the color compression is repeated again, the deterioration of the image quality of the blocks originally below the threshold value can be expected. Therefore, the threshold value to be reduced when the weighting is performed in advance is set to 8. A value smaller than 0 (for example, 7.0) is desirable.

An image color compression procedure according to this method is shown as a flowchart in FIG. In addition, the steps ST401 to ST405 in FIG.
Since the processing is the same as that of 101 to step ST105, description thereof will be omitted here.

In step ST406, it is determined whether the image quality deterioration parameter D of all blocks is equal to or less than the threshold value T ₁ . If so, it is considered that the desired image quality has been obtained, and the series of image color compression processing ends. If at least one block exceeds the threshold value T ₁ , the process proceeds to the next step ST407 and the process is continued. Here, as in the above example, the standard deviation is used as the image quality deterioration parameter D and 8.0 is used as the threshold value T ₁ .

In step ST407, a block exceeding a weighted threshold value T ₂ (for example, 7.0) smaller than the threshold value T ₁ is searched, and only the block exceeding this T ₂ is targeted, and the image quality of the block is searched. The appearance frequency of the color data is added based on the value obtained by subtracting the weighted threshold value T ₂ from the deterioration parameter D. Then, the process returns to step ST403 and the color compression process is repeated.

In each of the above embodiments, the standard deviation and the discontinuity evaluation value are given as parameters for obtaining the degree of image quality deterioration, but the content rate of skin color data may be used as a parameter. When humans judge image quality, the evaluation of skin color, which is most regarded in social life, has a great influence. Therefore, the false contour in which the deterioration of the skin color data appears ubiquitously in a part of the image causes the image quality evaluation of the color compressed image to be greatly deteriorated, and the block including the skin color data of a certain number of pixels or more is weighted. This makes it possible to remove the false contours that are noticeable.

In each of the above embodiments, the original image is described as 8 bits for each of R, G, and B, but it is obvious that the present invention can be applied even when the number of bits is other than 8 bits. The representative color is not limited to 256 colors. Furthermore, although the RGB space has been described in each of the above-described embodiments, the color compression according to the present invention may be performed on an original image represented in a YUV space or an HLS space as long as the space has color information for each pixel. You can

[0027]

As described above, according to the present invention, it is possible to automatically evaluate and remove false contours existing in a part of an image, and reduce the time required for image color compression.

[Brief description of drawings]

FIG. 1 is a flowchart showing a processing procedure of an image color compression method according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an example of an original image and its color-compressed image.

FIG. 3 is an explanatory diagram showing an example of block division of a color compressed image and an example of image quality evaluation data of each block according to an embodiment of the present invention.

FIG. 4 is a flowchart showing a processing procedure of an image color compression method according to another embodiment of the present invention.

FIG. 5 is an explanatory diagram showing a conventional image quality evaluation method.

[Explanation of symbols]

 201 False contour appearing in color compressed image 301 Block including false contour 501 Cursor for designating false contour

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁵ Identification code Internal reference number FI Technical display location G09G 5/06 9175-5G 5/36 9177-5G H04N 1/40 D 9068-5C 1/46 9068 -5C (72) Inventor, Ho Ho Ito, 292 Yoshida-cho, Totsuka-ku, Yokohama, Kanagawa Pref., Institute of Visual Media, Hitachi, Ltd. (72) Inventor, Takashi Takeuchi, 292, Yoshida-cho, Totsuka-ku, Yokohama, Kanagawa Prefecture Hitachi, Ltd. Visual Media Research Center

Claims (4)

[Claims] 1. A color data appearance frequency distribution of each pixel of an original image in a three-dimensional color space is obtained, color data having a high appearance frequency is extracted to determine a specific number of representative color data, and the original image of the original image is extracted. In the image color compression method for creating the image code data obtained by replacing the color data of each pixel with the representative color number of the representative color data, creating a color compressed image composed of the representative color data and the image code data, The color compressed image data is divided into a plurality of blocks, the color data of the original image and the color data of the color compressed image are compared in each block to obtain the degree of image quality deterioration, and the degree of image quality deterioration is determined for each block. The appearance frequency number of the color data is weighted, the representative color data is determined again, and the image quality evaluation and weighting for each block described above are performed until convergence to a desired image quality deterioration or Repeating a predetermined number of times that is determined because the image color compression method, which comprises carrying out color compression of the original image. 2. The color data according to claim 1, wherein the color data of the original image and the color data of the color compressed image in each block are compared, and the frequency of appearance of the color data is heavily weighted for a block having a large image quality deterioration. An image color compression method characterized by assigning a small weight to the frequency of appearance of color data for blocks with little image quality deterioration. 3. The color data according to claim 1, wherein the color data of the original image and the color data of the color compressed image in each of the blocks are compared, and the frequency of appearance of the color data is set only for a block exceeding a specific threshold value. An image color compression method characterized by weighting. 4. The frequency of appearance of color data of a block according to claim 1, when the number of pixels of skin color in a block exceeds a specific number in the process of obtaining the degree of image quality deterioration in each block. An image color compression method characterized by weighting numbers.