JPH0887256A - Color reduction method - Google Patents

Abstract

PURPOSE: To provide a natural, smooth and attractive image whose display and print are not rough by replacing a decomposed color of a pixel of coincidence to a comparison reference color becoming a comparison partner and replacing the decomposed color of the pixel of noncoincidence to the comparison reference color whose gradation is the nearest. CONSTITUTION: An RGB block control part 3 decomposes respective pixels of a full- color image to red, green and blue when the full-color image is inputted. Then, the number of use pixels are counted in gradation unit in subdivided blocks corresponding to the gradations of respective decomposed colors, and the number of use gradations in respective subdivided blocks are counted. Thereafter, the decomposed color of which use frequency is more is selected from the whole subdivided blocks. Then, the decomposed color of the selected gradation is decided as the comparison reference color, and the comparison reference color is compared with the decomposed colors of each pixel of full color image of a color reduction object successively. Then, the decomposed color of the pixel of which both coincide with each other is replaced to the comparison reference color becoming the comparison partner, and the decomposed color of the pixel of which both disagree is replaced to the comparison reference color of which gradation is the nearest, and a color reduced image 4 is generated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color reduction method for reducing the number of colors of a full color image output by an image output device such as a display device or a printing device.

[0002]

2. Description of the Related Art Conventionally, when outputting a full-color image with an image output device such as a display device or a printing device, there is a device that creates an image by dithering by extracting an average color from the full-color image. is there.

Dithering is a method in which different colors are set alternately in a certain special pattern and pseudo intermediate colors are created.

For example, when the display device can specify 256 arbitrary colors and can display 256 colors at the same time, the full color (for example, 167) is selected without selecting only the colors that are frequently used.
Average color is extracted from the image (70,000 colors or 260,000 colors),
This is a method to prevent color deviation.

In the case of this method, since the colors are extracted on average, if two adjacent colors are displayed,
It is displayed using the color whichever is closer.

Further, when there are a plurality of similar color portions in a large area in the image, the colors of these portions are filled with the same color. Therefore, even if the shape and the depth are represented by the approximate color, the display is flat.

In another case, for example, one dot is purple. However, when there is no purple in the 256 colors, if red and blue are placed next to each other by one dot, they will be mixed and appear purple to the human eye. That is, a plurality of colors can be used to make different colors appear, and the image can look good when viewed as a whole image. However, since the color changes dot by dot, the color of the image loses smoothness. Therefore, when the target object in the image is large, it is effective, but when it is small, the effect is weakened. Furthermore, when a display device having a low resolution is used, the roughness becomes conspicuous.

On the other hand, in a full-color image, when each of red (R), green (G), and blue (B) has 64 gradations, the number of expressible colors is 262,144 (= 64 × 64 × 64). )
It becomes a color.

Under such conditions, when the display device can simultaneously display an arbitrary color, for example, up to 256 colors at the same time, as a method for reducing the number of colors, 256 colors are used first in the order of frequency of use.
The color was extracted, and the remaining color was extracted first.
A method of changing to the closest color among the six colors can be considered.

With this method, although the image quality can be obtained to some extent, the remaining color is changed to an unexpected color because the gradation is approximated numerically, and the roughness is conspicuous around the image. It will not look good.

On the other hand, there has been proposed a color tone correction method as disclosed in JP-A-63-151267 or JP-A-63-151270.

The color tone correction method disclosed in Japanese Patent Laid-Open No. 63-151267 is a method of obtaining an image whose color tone has been corrected by using simple parameters.

The color tone correction method disclosed in Japanese Patent Laid-Open No. 63-151270 is a method in which a black detector for identifying black or a color close to black is provided, and these are treated as one black color.

[0014]

To reduce the number of colors in a full-color image, for example, a sunset with a lot of yellow-reds, a beach with a lot of blues, a person with a lot of reds, a snow mountain with a lot of whites, etc. As described above, in consideration of the tendency that the colors used in the image are concentrated in the same color system, it is necessary to reduce the colors so that the appearance does not feel roughness or unnaturalness.

However, in the conventional method for reducing the color of the full-color image by the dithering process,
There are problems such as a two-dimensional image and an unnatural image without smoothness, and there is a problem that it cannot be processed unless the area of a full-color image is large to some extent.

Further, in the color tone correction method disclosed in JP-A-63-151267 and JP-A-63-151270, since the color correction is performed using simple parameters, it depends on the parameters. Although it is possible to reduce the number of colors, the parameter is artificially set, so that there is a problem that the image becomes unnatural with respect to the original image.

The object of the present invention is, for example, 16.77 million colors or 2
Providing a color-reduction method that can reduce a full-color image consisting of an enormous number of 60,000 colors to 256 colors, etc., to a good-looking image that does not cause roughness or unnaturalness in the image quality It is to be.

[0018]

In order to achieve the above object, the present invention provides an RGB space having a logical three-dimensional structure composed of three lightness axes of red (R), green (G) and blue (B). The color of each pixel of the full-color image to be subjected to color reduction is divided into a plurality of subdivided blocks in a predetermined gradation unit, and the color of each pixel is red (R), green (G),
It is decomposed into blue (B), and the number of pixels used for each separated color is counted in gradation units in a divided block corresponding to the gradation of each separated color among the plurality of divided blocks, and
After counting the number of tones used in each subdivided block, the number of separated colors of the tones having a large number of used pixels from each subdivided block is selected by the number corresponding to the number of tones used in each subdivided block, The separation color of the selected gradation is defined as a comparison reference color, and the comparison reference color is sequentially compared with the separation color of each pixel of the full-color image to be reduced, and the separation color of the pixel where both match is the comparison partner. Replaced with the reference color
The separated color of the mismatched pixel is replaced with the comparison reference color having the closest gradation, and the color of the full-color image to be reduced is reduced to the number of the comparison reference colors.

[0019]

According to the present invention, an RG having a logical three-dimensional structure composed of three lightness axes of red (R), green (G) and blue (B).
Prepare B space.

The RGB space is for three-dimensionally expressing the brightness of red (R), green (G), and blue (B), and each axis is a gradation and the intersection of three brightness axes ( (Origin) is "black", vertices other than origin are the brightest red, green,
It represents blue. The intersection of all bright red, green and blue represents "white".

Then, the RGB space is divided into a plurality of subdivided blocks in units of a predetermined gradation. For example, 262,144 (= 64 ×, each of which has 64 gradations of red, green, and blue)
When handling a 64 × 64) full-color image, 64 gradations are divided into 8 gradations in units of 8 gradations, and 5 corresponding to the number of divisions.
It is subdivided into 12 (= 8 × 8 × 8) blocks.

In this way, the number of colors that can be represented by one subdivided block is 512 (= 8 × 8 × 8).

Therefore, the color of each pixel of the full-color image to be subjected to color reduction is decomposed into red (R), green (G), and blue (B), and the number of pixels used is determined for each of the separated colors of a plurality of subdivided blocks. The subdivision blocks corresponding to the gradation are counted in gradation units, and the number of gradations used in each subdivision block is counted. That is, how many colors the target full-color image uses in each of the 512 subdivided blocks in the RGB space and how many pixels use each color are counted.

After that, from each subdivided block, the separated colors of the gray scales having a large number of used pixels are selected by the number corresponding to the used gray scale number in each subdivided block.

As a result, a predetermined number of frequently used colors are extracted from the colors forming the full-color image.
For example, only 256 colors are extracted.

Next, the separated color of the selected gradation is defined as a comparison reference color, and the comparison reference color and the separated color of each pixel of the full-color image to be subtracted are sequentially compared, and the pixels of which the both match are separated. The color is replaced with the comparison reference color which is the comparison partner, and the separated color of the mismatched pixel is replaced with the comparison reference color having the closest gradation.

As a result, the color of the full-color image to be reduced is reduced to the number of comparison reference colors.

In this case, since the number of separated gradation colors having a large number of pixels used from each subdivided block is selected according to the number of used gradations in each subdivided block, for example, a full color on a beach with many blues. Since many blue-based comparison reference colors are selected in the image, the image becomes a natural image that reflects the color characteristics of the target full-color image.

[0029]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a functional block of an image processing apparatus for implementing the present invention. An image scanner 1 for inputting a full-color image, a red (R) for counting the frequency of use of each color of the input full-color image, A logically structured RG consisting of three lightness axes of green (G) and blue (B)
B block (space) 2, RGB block control unit 3 for reducing the color of a full-color image using this RGB block, display control unit 5 for displaying the reduced color-reduced image 4 on the display device 6, and similarly reduced color. A print control unit 7 is provided to print out the reduced color image 4 from the printing device 8.

Hereinafter, the brightness of red, green, and blue is 64 gradations, and the full-color image is 262,144 (= 64 × 64 ×).
64) and the target number of colors (color-reduced image 4) will be described with an example of 256 colors.

The RGB block 2 is for expressing the brightness of red (R), green (G), and blue (B) in a three-dimensional structure, and each axis is a gradation and has three brightness axes. The intersection point (origin) of is black, and the vertices other than the origin represent the brightest red, green, and blue, respectively. The intersection of all bright red, green and blue represents "white".

In the present invention, as shown in the enlarged view of FIG.
512 (= 8 × 8 ×) of the RGB block 2 in units of 8 gradations
8) Divide into subdivided blocks 20.

In this way, the number of colors that can be represented by one subdivided block is 512 (= 8 × 8 × 8).

Since the subdivision block 20 is for counting the frequency of use of each color of the full-color image, it is specifically composed of a gradation unit counter using a memory.

First, in the case of reducing the full-color image of the color-reduction target, the target full-color image is input from the image scanner 1.

When the full-color image is input, the RGB block control unit 3 decomposes the color of each pixel of the full-color image into red (R), green (G), and blue (B), and divides into 512 pieces. In the subdivided block corresponding to the gradation of each separated color in the block 20, the number of used pixels n is counted in gradation units, and the number of used gradations m in each subdivided block 20 is counted. That is, the target full-color image is the 512 subdivided blocks 2 in the RGB space.
The number of colors used in 0 and the number of times each color is used are counted.

After that, from the entire subdivided block 20, 256 separated colors having gradations that are frequently used are selected.

In this case, how many separation colors are selected from which subdivided block 20 is selected for each subdivided block 2 among the number of colors and the number of pixels of the entire target full-color image.
Selection is made according to the used color number ratio of 0 and the used dot number ratio.

The used color number ratio is the number of used colors (= m) / the total number of colors (= 262,144) in the subdivided block, and the used dot number ratio is the number of used subdivided blocks. The number of used pixels (= n) / the total number of pixels.

Concretely, the number of gradation separations having a large number of used pixels n is sequentially selected by the number corresponding to the number m of gradations used in each subdivision block 20, and finally 256 separated colors are obtained. Select.

As a result, a predetermined number of frequently used colors are selected from the colors forming the full-color image.
For example, only 256 colors are selected, and 260,000 colors are narrowed down to 256 colors.

Next, the separated color of the selected 256 gradations is defined as a comparison reference color, and the comparison reference color and the separated color of each pixel of the full-color image to be subtracted are sequentially compared, and both match. The separated color of the pixel is replaced with the comparative reference color that is the comparison partner, and the separated color of the non-matching pixel is replaced with the comparative reference color having the closest gradation, thereby generating the reduced color image 4.

Now, it is assumed that the number of used pixels n in one gradation unit counted in one subdivision block 20 has a distribution as shown by the numerical values in the frame of FIG.

Although the subdivided block 20 actually has a cubic structure, it is described here as a planar structure instead of a cubic structure in order to make the description easy to understand. The vertical axis represents red brightness and the horizontal axis represents green brightness, and the brightness increases from the bottom to the top and from left to right in the figure. It becomes bright red at the top of the figure and bright green at the right. Going diagonally to the right, it becomes a bright yellow mixture of red and green.

The numerical value in the subdivided block is the number n of pixels used for each gradation corresponding to the brightness of red and green.

In this example, the color having the brightness of "0" for the red lightness and "0" for the green lightness is "35" pixels. In addition, the color having the brightness at the position where the red lightness is "5" and the green lightness is "5" is the most in 95 pixels.

From here, if selected in descending order of frequency of use,
The order is "95", "90", "90", "85" .... Actually, since the subdivision block 20 is a cube, 51
Two pieces will be arranged in descending order, and an arbitrary number of them will be extracted by the above-described indexing method.

When 10 gradations are extracted by the descending extraction method, a gradation as indicated by reference numeral 10 in FIG. 4 is selected. In this case, the lightness is concentrated and extracted. If the darker color cannot be extracted and the 256 colors have not been matched, the colors may be separated when the approximate color is selected. In addition, since the colors are concentrated, only those whose color differences are almost the same are extracted.

In the selection method using the RGB block 2,
Extraction is performed in descending order, but the top, bottom, left, and right (including the front and rear in the case of a solid) of the extracted colors are already extracted, and they are not extracted thereafter.

10 by the selection method using the RGB block 2.
When the individual pieces are extracted, the gradation as indicated by reference numeral 11 in FIG. 4 is selected.

By the descending extraction method, the colors can be extracted in a dispersed manner, and the colors are not separated even when the approximate color is selected.

The display control unit 5 uses the thus generated color-reduced image 4 to display it on the display device 6 such as a graphic display.

Further, the print controller 7 causes the printing device 8 capable of printing the color-reduced image 4 in color.

As described above, the color of the full-color image to be reduced is reduced to the number of 256 comparison reference colors.

In this case, since the separation colors of the gradations having a large number of used pixels are selected from each of the subdivided blocks 20 by the number corresponding to the number of the used gradations in each of the subdivided blocks 20, for example, a beach with a lot of blue colors In the full-color image, since many blue-based comparison reference colors are selected, the image becomes a natural image that reflects the color characteristics of the target full-color image.

[0057]

As described above, according to the present invention, a plurality of RGB spaces having a logical three-dimensional structure composed of three brightness axes of red (R), green (G), and blue (B) are provided in units of a predetermined gradation. The color of each pixel of the full-color image to be subjected to color reduction is divided into red (R), green (G), and blue (B) in advance and divided into subdivided blocks. The number of used pixels is counted for each gradation in the subdivision block corresponding to the gradation, the number of used gradations in each subdivision block is counted, and then the gradation having the large number of used pixels is decomposed from each subdivision block. The number of colors is selected according to the number of gradations used in each subdivision block, the separated color of the selected gradations is defined as a comparison reference color, and the comparison reference color and each pixel of the full-color image to be reduced Compare the separated colors one by one, and The color separation is replaced with the comparison reference color that is the comparison partner, the separated color of the mismatched pixel is replaced with the comparison reference color with the closest gradation, and the color of the full-color image to be reduced is reduced to the number of the comparison reference colors. As a result, the display and printing of the image are not rough, and an image with a natural and smooth appearance can be obtained.

[Brief description of drawings]

FIG. 1 is a functional block diagram showing an embodiment of an image processing apparatus for realizing the present invention.

FIG. 2 is an enlarged view of an RGB block (space).

FIG. 3 is an explanatory diagram showing a numerical example of the number of pixels used in one subdivided block.

FIG. 4 is an explanatory diagram showing an example of a process of selecting a comparison reference color used for color reduction.

FIG. 5 is an explanatory diagram showing another example of a process of selecting a comparison reference color used for color reduction.

[Explanation of symbols]

1 ... Image scanner, 2 ... RGB block, 3 ... RG
B block control unit, 4 ... reduced color image, 5 ... display control unit,
6 ... Display device, 7 ... Print control unit, 8 ... Printing device, 20 ...
Subdivision block.

 ─────────────────────────────────────────────────── ─── Continued Front Page (72) Inventor Masanori Sugie 6-81, Onoe-cho, Naka-ku, Yokohama-shi, Kanagawa Hitachi Software Engineering Co., Ltd. (72) Inventor Ryuji Higada 5030 Totsuka-cho, Totsuka-ku, Yokohama-shi, Kanagawa Ceremony Company Hitachi Software Development Division

Claims (1)

[Claims] 1. A logically three-dimensional RGB space composed of three lightness axes of red (R), green (G), and blue (B) is divided into a plurality of subdivided blocks in predetermined gradation units. , The color of each pixel of the full-color image to be reduced is separated into red (R), green (G), and blue (B), and the number of pixels used for each separated color is the separated color of the plurality of subdivided blocks. While counting in gradation units in the subdivision block corresponding to the gradation of, the number of gradations used in each subdivision block is counted, and then the separated color of the gradation in which the number of pixels used is large from each subdivision block, A number corresponding to the number of used gradations in each subdivision block is selected, and the separated color of the selected gradation is defined as a comparison reference color, and the comparison reference color and the separated color of each pixel of the full-color image to be reduced Are sequentially compared, and the separated colors of the pixels where they match are compared. It is characterized in that the color of the full-color image to be color-reduced is reduced to the number of the comparison reference colors by replacing it with the comparison reference color that is the partner And the color reduction method.