JP2775748B2 - Method and apparatus for expressing limited color of color image - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Application Field) The present invention is intended to display a digitized natural color image on a color image display device which can simultaneously display only a limited number of colors, without any discomfort. The present invention relates to a method and an apparatus for expressing in a limited color.

(Prior Art) A color image is usually represented by three components of R (red), G (green), and B (blue). To display a natural color image,
Each component is digitized to at least 5 bits, usually about 8 bits. If each component is digitized to 8 bits, each pixel requires 24 bits of memory. FIG. 4 shows such a display device, in which each pixel
The R, G, B values stored in the 24-bit image memory 1 are converted into voltages by the respective D / A converters 8 and displayed on the CRT 9.

Meanwhile, CAD (Computer Aided Design)
In the field of graphics and graphics, a color display device capable of simultaneously displaying only a limited number of colors is often used. This example will be described with reference to FIG. The image memory 10 has a capacity of 8 bits for each pixel and takes a value of 0 to 255. This pixel value indicates a code of a color displayed in each pixel. The pixel values are input to the color map 7 when displayed. The color map 7 is a high-speed memory having addresses from 0 to 255, and reads a total of 24 bits, each of 8 bits indicating the R, G, and B values of the accessed address. R, G and B are converted into voltages by the respective D / A converters 8 and displayed on the CRT 9. The display device of the type shown in FIG. 4 is hereinafter referred to as a full-color display for convenience, and the display device of the type shown in FIG. 5 is referred to as a color map display. The full-color display, it is possible to choose freely the color from 2 ²⁴ ≒ 1600 million colors in each pixel.
On the other hand, even with a color map display, the output of the color map is 24 bits, so that colors can be selected from 16 million colors. However, since the capacity of each pixel of the image memory 10 is 8 bits, the colors that can be displayed simultaneously in an image can be selected. Will be 256 colors. Normally, a natural color image cannot be displayed with a limited number of colors such as 256 colors without deterioration, but there is a strong demand to make this possible.

The requirement is to simply sample 256 colors uniformly in a color area that can be displayed on a color display, create a representative color set, and then, for each pixel of the color image, use the colors included in the representative color set, The color of the pixel can be filled by choosing and assigning the closest color. This method is the sixth
This will be described with reference to the drawings. The area that can be displayed on the display is represented by a cube in the RGB space, but is drawn in the RG two-dimensional space for simplicity in FIG. Here, the representative color sets (○) are selected at equal intervals. When a certain pixel in the image has a color indicated by a cross 101, the color of this pixel is represented by a representative color 102 closest to this. By the way, when the representative colors are sampled at equal intervals in the RGB space, the distance between two points in the RGB space is not proportional to the visually perceived color difference between the colors.
The fidelity of color reproduction differs depending on the color of the image. In order to make this uniform, there is known a color space using a uniform color space in which the distance between two points in the space is approximately proportional to the color difference (Information Processing, Vol. 27, No. 1) , 1986, 11
~ 20 pages).

However, with the above method, good quality images
When about six colors are used, the color difference between the representative colors is large,
In many cases, only unsatisfactory images can be displayed. The above-mentioned literature also describes a known method of selecting a representative color adapted to the color distribution of each image instead of the fixed 256 colors and displaying the image naturally. However, in this case, considerable computer processing is required to select a representative color suitable for the color distribution of each image.

In order to solve this, a method using a so-called dither method is known. In this method, the light emission intensities of the primary colors are spatially averaged so as to be faithfully reproduced.

For the 8-bit data of R, for example, 16 steps of R = 0, 16, 32,..., 240 are selected as representative colors, and the upper 4 bits of R are displayed as a part of a color code. When the closest representative color is allocated, all 16 levels of R = 8 to 23 are R
= 16. As shown in FIG. 7, two neighboring 2 × 2 pixels are R = 0, and the remaining two pixels are R = 0.
= 16 on average, R =
Eight colors can be visually displayed with almost no image deterioration. This technique can be realized with the configuration as shown in FIG. Each pixel value read from the image memory 1 is divided into upper 4 bits 11 and lower 4 bits 12. The threshold image memory 13 stores 2 × 2 threshold images.
The threshold image memory 13 is addressed by the lower 1 bit of the X and Y coordinate signals that address the image memory 1 (not shown), and the lower 4 bits of the pixel value are compared by the comparator 14 with the pixel value of the threshold image. When the lower 4 bits are smaller, 0 is set, and when the lower 4 bits are larger, 1 is set.
Is output. The adder 15 adds the output of the comparator 14 to the upper 4 bits and sets the emission intensity of the display as the R component of the code indicating the representative color.

(Problems to be Solved by the Present Invention) However, this conventional dither method cannot be applied to a case where representative colors are uniformly selected in a uniform color space in which uniformity of color differences between representative colors is good. This will be described with reference to FIG. The figure shows v ^* = in L ^* u ^* v space, which is one of the uniform color spaces.
Although the displayable area 16 of the color display is schematically drawn on the cross section of 0, since this shape does not become a rectangle (that is, a rectangular parallelepiped in a three-dimensional space), for example, L ^* =
L ₁ ^{*, v} display possible number of colors in * = _{^{0, L * = L 2 *,}} v *
= 0 is not necessarily equal to the number of displayable colors (the ninth
Figure x mark). Therefore, the pixel value (L ^* , u ^* ) is a black circle 103
Will be erroneously requested for a grid position (× 104) that cannot be displayed in order to express

The present invention has been made to solve the above problems, and to provide a method and apparatus for expressing a natural color image expressed in a uniform color space by a predetermined representative color and a dither method without discomfort regardless of the color of the image. It is an object.

(Means for Solving the Problems) In the first invention, a color image is represented by components of a uniform color space, and a pixel value of a corresponding dither pattern is added to each pixel of the color image to obtain a modulation pixel value. Obtaining the modulated pixel value
A limited color expression method of a color image expressed by the closest representative color of a representative color set sampled at equal intervals in the displayable area of the uniform color space.

According to a second aspect of the present invention, there is provided the method for expressing a limited color of a color image according to the first aspect, wherein the uniform color space is a modified uniform color space in which the lightness direction of the uniform color space is extended about four times. It is.

According to a third aspect of the present invention, there is provided a dither pattern storing means for storing a pixel value of a dither pattern to be added for each pixel of a color image represented by three components of a uniform color space, and for each pixel of the color image An adder group for adding a pixel value of a corresponding dither pattern for each component to obtain a modulated pixel value, and a latest color determining means for selecting a representative color closest to the color represented by the modulated pixel value and outputting a color code thereof , A limited color expression device for a color image.

According to a fourth aspect of the present invention, in the third aspect of the present invention, in the limited color expression device for a color image, the limited color expression device for a color image is a transformed uniform color space in which the lightness direction of the uniform color space is extended about four times. It is.

(Operation) The operation of the present invention will be described with reference to FIG. 10 and FIG. In order to prevent the color difference between the representative colors from largely changing depending on the colors, the representative colors are in a displayable area of the L ^* u ^* v ^* space, which is one of the uniform color spaces recommended by the CIE (International Commission on Illumination). For example, 256 colors are selected uniformly. This can be similarly performed in a space having similar properties such as an L ^* a ^* b ^* space.

Assuming that R, G, and B of each pixel of the image conform to the NTSC television system, R, G, and B are expressed by the equations (1) and (2) via the XYZ system recommended in CIE-1931. Converted to L ^* u ^* v ^* .

To add dither processing to this, a dither pattern corresponding to the threshold image in FIG. 8 is added to each component. For example,
When the lattice spacing is 16 as in the condition of FIG.
^{For *} , the dither pattern of FIG. 10 (a) is added, and for u ^* , the dither pattern of FIG. 10 (b) is added while being spatially shifted to each pixel. Then, L ^* = which was originally the color of the black circle position 105 in FIG.
The pixel value of 8 varies at the Δ position, and the pixels of 106 and 107 are
At L ^* = 16, the pixels at 108 and 109 are closest to L ^* = 0. Therefore, at the time of display, if the representative color of the closest color to the color indicated by the value obtained by adding the pixel value of the dither pattern is selected, the natural color image is approximated by the spatially averaged color. Can be displayed.

In this way, even when the representative colors shown in FIG. 9 are not distributed in a rectangular parallelepiped in space, the best result can be obtained by selecting and displaying the representative color that is three-dimensionally closest. .

To optimize the color reproduction in both the brightness direction and the chromaticity direction, the above processing is appropriate.However, when expressing a natural color image with limited colors, the difference in the lightness direction is the same even if the color difference is the same. It is known that a false contour is conspicuous. Also,
It is also known that the difference in the chromaticity direction is not very noticeable unless compared with the original image. It is appropriate to sample the color about four times denser than in (1). In this case, it is better to newly define L 'as in equation (3), uniformly sample in L'u ^* v ^* space and select a representative color, and perform the above processing in this space.

L '= 4L ^* (3) (Example) An example of the present invention will be described with reference to FIG. The image memory 1 is scanned by the X and Y coordinates generated by the image scanning means 2, and the (R, G, B) values of each pixel value are converted by the color conversion means 3.
Is output to The color conversion means 3 converts the pixel value into (L ^* , u ^* , v ^* ) according to the above equations (1) and (2). The dither pattern storage unit 4 is an image of m × n (m and n are natural numbers), and is also scanned by the image scanning unit 2. Where X,
Both Y-coordinates are repeatedly scanned in the surrounding areas m and n.
The dither pattern has m × n levels and the representative color is L ^*
When the color difference sampled in the u ^* v ^* space is d,
There is one level for each color difference of d / (m + n), and the level is created so that the average is 0. For example, in the case of d = 16 and m = n = 2 described above, the data is created as shown in FIG. 10 (a). The dither pattern is added for the three values of L ^* , u ^* , and v ^* . In this case, the shape of the dither pattern can be changed. In the above example, Fig. 10 (a)
When the dither pattern is used for the values of the three components at the same time, a large color difference always occurs in the display image corresponding to the pair (−6, 6) in the first row, and the dither pattern corresponds to the pair (2, −2) in the second row. Since a color difference is not likely to occur in a display pixel, a periodic pattern is easily conspicuous. Therefore, when the pattern of FIG. 10A is used for L ^* , u ^*
For, FIG (b), ^{v *} for the drawing (c)
Can be used to adjust the average color difference between the pixels so that the pattern does not stand out.

The adder group 5 adds the pixel values of the dither pattern for each of L ^* , u ^* , and v ^* . The recent color determination means 6
(L ^* , u ^* , v ^* ) value (L ^* ', u) which is the output of adder group 5
^* ′, V ^* ′), the closest color among the representative colors is selected, and the representative color number is output. The output representative color number is written to the corresponding pixel of the image memory 10 as the pixel value of the color code image to be displayed on the color map display.

The scope of the present invention has been described so far. However, when this is displayed on a color map display, the following is performed.

The color map 7 stores representative colors as (R, G, B) values. The image memory 10 is scanned and the representative colors are output from each pixel. , B) values are accessed and applied to each D / A converter 8 and the color is displayed on the CRT 9. In this way, the natural color image is approximately represented by the limited color including dither.

As described above, in the case where the image is not compared with the original image, in order to reduce the false contour, the calculation in the color conversion means 3 is performed as L ^* , u ^* , v as shown in Expression (3). ^By using L ', u ^* , v ^* instead of ^* , it is possible to obtain a limited color display including dither with inconspicuous outlines in exactly the same manner. However, in this case, since the volume of the displayable area is quadrupled in the brightness direction, in order to make the number of representative colors the same, the interval of sampling the representative colors needs to be approximately 4 ^1/3 d from d.

The above-described recent color determination means 6 can be configured by a known technique. In order to accurately determine the closest color, it is necessary to compare the distances of all the colors of each pixel with all the representative colors, but good results are often obtained by an approximate method. One example is the second
This will be described with reference to FIG. 3 and FIG. For the sake of simplicity, it is assumed that the representative colors are sampled at equal intervals on a two-dimensional u ^* -v ^* plane instead of a three-dimensional plane, and are indicated by a circle in FIG. Here, the representative colors are eight colors. In the first stage, the color of each pixel is determined to be on either side of the four-color area divided by the solid line of the boundary 601. That is, the pixel value may be projected in a direction perpendicular to the boundary 601 and compared with the position of the boundary. In practice, the color space is three-dimensional, the boundaries are represented by a plane, and the pixel values are projected onto a normal to this plane. Specifically, the coefficient
A, B, C, and D are obtained, and it is determined which side is on the basis of the sign of the value of A ^* L ^* + Bu ^* + C ^* v ^* + D (4).

In the second stage, two more areas are determined in the area determined in the first stage.
It is determined on which side of the area for each color. It again projects pixel values in a direction perpendicular to the boundary 602 or 603 and compares it with the position of the boundary. The determination is made by the same calculation as in equation (4). In the third stage, the pixel values are associated with any of the eight representative colors in the same manner.
If a binary value of 0 or 1 is made to correspond to the determination result of the area at each stage, the representative color is coded by a 3-digit binary number.

FIG. 3 shows an embodiment of the latest color judging means 6 for performing the above processing. The pixel value (L ^* , u ^* , v ^* ) is added to the projection calculation means 61.
Equation (4) is operated according to, and the comparing means 67 determines whether the operation result is positive or negative, and outputs 0 or 1 as a result of the first stage. Coefficient storage means 6 as coefficients A, B, C, D
Use the one stored in 4. The projection calculation means 62 in the second stage reads out the coefficients A, B, C, and D from the coefficient storage means 65 by using the code of the result in the first step, and performs the operation of equation (4). For example, if the result of the first stage is 0, the coefficient read is for the boundary 602, and if the result is 1, the coefficient is for the boundary 603. The second digit of the code is output by the comparing means 68 as a result of the second step. The projection calculation means 63 in the third stage includes the same coefficients A,
B, C, and D are read from the coefficient storage means 66 using the two-digit code obtained as a result of the first and second steps, and the operation of equation (4) is performed. The third digit of the code is output by the comparing means 69, and a three-digit code indicating the representative color is obtained. If this process is repeated up to eight levels, a recent color judging means for displaying 256 colors (8 digits) is realized.

(Effects of the Invention) When a color image is expressed by the method and apparatus described above, a natural color image originally displayed on a full color display can be displayed without a sense of incongruity even with a color map display that can simultaneously display only a small number of limited colors. be able to. In addition, when storing color images, image data that originally required three bytes of R, G, and B can be reduced to one byte, for example, so that the capacity of the storage device and the amount of transfer data can be reduced. It has great industrial and economic effects. In this case, although it takes a long processing time to reduce the number of colors according to the present invention, no decoding processing is required to display the compressed image, which is very effective. In addition, since the set of representative colors is fixed for every image, there is also an effect that the method is executed at a higher speed than the method of selecting a representative color for each color distribution of each image.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is an explanatory diagram of the operation of the latest color judgment means, FIG. 3 is a block diagram showing an embodiment of the latest color judgment means, and FIG. FIG. 5 is a block diagram showing the configuration of a color map display, FIGS. 6 to 8 are explanatory diagrams of a conventional dither method, and FIG. 9 is a diagram showing a conventional color image expressed in a uniform color space. FIG. 10 is an explanatory diagram showing that the dither method cannot be appropriately applied, FIG. 10 is an example of a pixel value of a dither pattern,
FIG. 11 is an explanatory diagram showing the effect of the dither method used in the present invention. DESCRIPTION OF SYMBOLS 1 ... Image memory, 2 ... Image scanning means, 3 ... Color conversion means, 4 ... Dither pattern storage means, 5 ... Adder group, 6 ... Recent color judgment means, 7 ... Color map, 8 …
... D / A converter, 9 ... CRT, 10 ... Image memory, 13 ... Threshold image memory, 14 ... Comparator, 15 ... Adder.

Claims (4)

(57) [Claims] 1. A color image is represented by components of a uniform color space.
For each pixel of the color image, a pixel value of the corresponding dither pattern is added to obtain a modulated pixel value, and the modulated pixel value is sampled at equal intervals in the displayable area of the uniform color space. A limited color expression method for a color image expressed by the closest representative color. 2. A method according to claim 1, wherein the uniform color space is a modified uniform color space in which the lightness direction of the uniform color space is extended by about four times. 3. A dither pattern storage means for storing pixel values of a dither pattern to be added for each pixel of a color image represented by three components of a uniform color space; An adder group for adding a pixel value of a corresponding dither pattern for each component to obtain a modulation pixel value, and a latest color determination unit that selects a representative color closest to the color represented by the modulation pixel value and outputs a color code thereof; Device for expressing a limited color of a color image. 4. A color image limited color expression apparatus according to claim 3, wherein the uniform color space is set to a lightness direction of about 4
A limited color expression device for a color image in a transformed uniform color space that has been doubled.