Color Image Separation and Hanging Technology

1 Introduction Color printing often uses CMYK4 color overlay, relying on Qingyang 1999-02-12 to receive red, yellow, and black pigments to produce a color image. These four basic colors combine to present up to eight colors: red, green, blue, cyan, magenta, yellow, black, and white, and a color image contains far more than eight colors. Units with large-scale professional color publishing systems, such as color production centers or printing plants, use color image processors to color-code color images into 4-color grayscale images, and then raster images (RIP) to grayscale images. Using the size of the dot or dot density to reflect the continuous tone of the grayscale image, the final four-color printing is performed using a laser photocopying mechanism to produce four-color films. Most color image makers do not have professional image processing technology and expensive image processing equipment. They hope that the color image can be arranged through a interface software to achieve color image color separation, color correction and hanging network, and finally to the printing plant Plate printing, which is more economical in terms of system cost and flexibility. Color image separation and hooking are the main problems to be solved by this interface software.

2 color image separation technology 2.1 RGB color model and CMYK color model In order to facilitate computer processing of color, often use the color model to describe the color. There are many kinds of color models that describe colors. Among them, RGB color models are used for CRT color image display, and CMYK color models are used for color image printing.
RGB color model RGB color model is also called the additive color model. The color comes from the superposition of different brightnesses of the three basic colors of red, green, and blue. Therefore, the additive color model is called. It is mainly used to describe the color of light emitting devices, such as displays, televisions, scanners and other devices. In the model, the red, green, and blue light colors are measured with 256 tone values. Each channel assigns a numerical value to describe its tone, and the combination of the three light values ​​of different tone values ​​creates a rich color pattern. Color space, such as: each channel tone value is 255, the combination can produce white light; red channel bit 255, the rest is 0, you can simulate the effect of pure red; each channel tone value is 0, then Pixels are black; pixels have different shades of grey when the gradation values ​​are the same for each channel.
· CMYK color model CMYK color model is also called subtractive color model. The colors come from the three basic colors of cyan, magenta, and yellow. These three kinds of primary colors absorb some colors from the white light on the irradiated paper, thereby changing the light wave to produce color, that is, from white light. Subtracting some colors to produce color, it is called a subtractive model. It is mainly applied to the occasion where the physical substances such as printing ink and toner produce color, such as the field of color printing. In the model, each pixel value of a color image is measured by the percentage of cyan, magenta, yellow, and black inks, the percentage of inks of light-color pixels is lower, and the percentage of inks of deep-color pixels is higher, and the case of no ink is white. .
In the CMY color mode, theoretically, white paper reflects 100% of incident light, and mixing CMY3 100% colors absorbs all light and produces black. In the actual printing, the paper always absorbs some light, and the cyan, magenta, and yellow primary inks inevitably contain some impurities. Therefore, the black formed by 100% of the 3 primary color combinations tends to appear turbid gray, and the blackness is not enough. To compensate for this defect, A black pigment, K-color, was added to the print and this was called the CMYK model.
· The relationship between RGB and CMYK color models The RGB and CMYK color models look very different, and in essence they are complementary. The color wheel can be used to describe this relationship, as shown in Figure 1. The colors on the color wheel are arranged with each other, and any color can be combined with the colors in the adjacent opposite color models. The color in the inverse color model that does not contribute to the formation of a color is called the complementary color of this color. It can be known from the color wheel that cyan, pink, and yellow are complementary colors of red, green, and blue, respectively. This relationship between the RGB color model and the CMYK color model is the basis for the conversion between the two models and lays a theoretical foundation for the color image color separation technology.

figure 1

2.2 Color Image Commonly used color images are images of the RGB color model. Color separation is the decomposition of color images into cyan, magenta, yellow and black color grayscale maps, so color separation technology is the first to convert various colors in color images from RGB color model to CMYK color model, and then color The image is stored as a grayscale image of four colors of cyan, magenta, yellow, and black. It can be seen that the key to color separation technology is the color space conversion technology.
Based on the above relationship between the RGB color model and the CMYK color model, we obtain the content ratios of C, M, Y, and K required when a certain color on the original image is produced by the CMYK ink combination. The two models have the following relationships: C=F-R; M=F-G; Y=F-B where F is the full-color value. For any color C, M, Y, if its component is not 0, there is a gray component whose size is min(C,M,Y). In the actual printing, black pigment is added in order to make up for the defect that black formed by subtractive primary colors is not black enough. At the same time, the use of black pigment can also save a large number of color pigments, directly with the black pigment to form a different gray level, instead of subtractive 3 primary color combination. In this way, after the introduction of the black pigment, there will be redundant components of blue, pink and yellow. The content of cyan, pink and yellow in the actual printing must be reduced by the content of the portion replaced by the black pigment. Therefore, the conversion relationship between the RGB color model and the CMYK color model can be illustrated in Figure 2:

figure 2

Some explanations are needed for this conversion relationship: (1) The gamut of the RGB color model has a larger color gamut than the CMYK color model. Some color combinations of the RGB color space cannot be represented by the colors of the CMYK color space. When the conversion is performed, these colors will be cut off and only converted to similar CMYK colors as much as possible, so there is a problem of one-time conversion. (2) The conversion formula of Figure 2 is only implemented under ideal conditions. In practical applications, due to factors such as the characteristics and composition of pigments, the effect of printing the image according to the above formula is difficult to meet the requirements of the application, so the conversion factor of the color space must be corrected to ensure the output of the color image. quality.

3 A color image hanging method 3.1 Hanging net basic image The color separated image is 4 gray scales with continuous tone. In 4 color printing, only one type of ink can be used for each printing, and the ink The concentration stays the same. In order to obtain a continuous tone at the time of printing, it is necessary to perform a netting process on the grayscale image. Hanging nets, also known as screening, is the process of breaking down continuous-tone images into dots. After adding the screen, the size and density of the dots reflect the depth of the actual color of the image. Based on human visual effects, when viewing an image from a close distance, the dots and their surrounding space create continuous tone artifacts. Larger dots look darker, smaller dots look brighter, dots densely look like Dark, sparsely populated areas look bright. There are many ways to hang the net. According to the different positions of the graphics, the nets can be divided into front-end hanging nets and back-end hanging nets; according to the method of forming net points, they can be divided into amplitude modulation (AM) and frequency modulation (FM). method. The front-end hanging net is also called software hanging net. It is to do the netting processing before the image is output in the layout, and then the image data after the netting processing is stored on the disk for the printing output. The characteristics of this method of hanging nets are: slow processing speed, occupying a large amount of disk space, but strong flexibility, easy to upgrade and change. The back-end hanging net is also called hardware hanging net, which is a method for high-speed netting of images by the RIP of the grid processor while the images are being output. The characteristics of this method are: fast processing, saving disk space, but it needs RIP support, is not easy to upgrade, and has poor flexibility. AM webbing is a half-tone method of printing by changing the size of the printing dots during printing. The dots are dark in the large dots and bright in the dots. In this method, moire is generated due to the interference of the grid pattern. This is a traditional hanging net technology. FM hooking is to print the dots of the same size in a random pattern, and the halftone method is implemented by changing the sparseness of the dots. The dark spots in many locations are dark and the places with few spots are bright. With this method, there are no rules for the placement of dots, and the print will not form a certain line and no interference pattern will be produced.
3.2 Error dispersion method The error dispersion method is a method of front-end frequency modulation hanging net, which is based on the error method. The error method compares the grayscale value of each image point of the original image with the threshold value. Image points larger than the threshold value are recorded as white dots, and image dots smaller than the threshold value are marked as black dots. With this method, a grayscale image having a continuous tone can be halftoned, so that the resulting image has a black and white contrast that is too obvious and the effect is not good. The error dispersion method generates a halftone point by comparing the gray value of the image point with a threshold value, and at the same time, spreads the error between the gray value of the image point and the threshold value to the image point around the image point to make the halftone error of the point. The performance is not obvious in the final result. For a 256-grayscale image, the threshold is 256/2=127, and there is a pixel with a grayscale of 150. After comparison, this point should be recorded as white, but in fact this point is not really white. The difference in gray level between white and white is 23, and the error 23 is distributed to the surrounding image points of the point in a certain way, so that the error has little effect on the output result. There are several ways to distribute the error to the surrounding points. Here are some commonly used error dispersion algorithms:
• Floyd-Steinberg filter algorithm X7351 where X represents the gray value of a point in the image. The algorithm first compares the gray value of the X-pixel to the threshold. The image is recorded as 1 or 0, ie white or black. Then the error is calculated, the error is assigned to the surrounding points, and the gray value of the surrounding points is modified. The filtering algorithm adds 7/16 of the error to the first pixel on the right side of X. The error 3/16 is added to the first point on the left of the next row. The error 5/16 is added to the next row. On the image point, 1/16 of the error is added to the first point on the right side of the next line, which spreads the X-pixel error to the surrounding image points. This process is repeated to perform halftoning and correction of the gray value for each image point in the image, and finally obtain a halftone image reflecting the hierarchical relationship of the original image. This method is theoretically very good and can reflect the original graph's hierarchical relationship and color well. However, if we make more points for error dispersion, the effect will be better. Therefore, we propose filtering that can involve many points. Device.
• Stucki Filtering Algorithm This algorithm further improves the Floyd algorithm. Since it involves more points, the output image is better, but it requires a lot of calculations, so the data processing speed is slow. X842484212421 Similarly, X is the grayscale value of the pixel. After the halftone is compared with the threshold, 8/42 of the error is added to the gray value of the first pixel on the left side of X, and 4/42 is added to the left of X. The gray value of the second image point, 2/42 is added to the gray value of the second pixel on the left of the next line, and so on until the error of 1/42 is added to the right of the second line of X. The gray value of the second image point. Similarly, for each image point in the image, such halftone and error correction processing is performed, and a better halftone output image can be obtained. The disadvantage of this filter is that, as mentioned above, the operation speed is slow and a large number of integer division and multiplication operations are required.
Burks filtering algorithm This algorithm is a compromise between the running speed and the amount of data for the two algorithms mentioned above. It involves a lot of pixels and only simple shift and addition and subtraction operations. The error distribution method it adopts is as follows: X8424842 is the same, X is the gray value of the image point. After the halftone processing is compared with the threshold value, a certain number of 32 points of the error is added to the corresponding image points around X, Each image point in the image.
For any of the above algorithms, in the case of error dispersion, if one pixel is always processed from one scan line to the next from left to right, an error of one scan line is simply added to the next line, thereby causing Accumulation of errors, manifested in the output graphics is a tendency for the image to drive away. So when scanning, often using 'S'-shaped scanning, that is, if the odd lines from left to right scanning, even lines