TW436712B - Processing method of halftone image - Google Patents

Abstract

The present invention provides a processing method of halftone image, wherein a halftone image technology capable of being used in fields of displaying, printing or publishing digital image is disclosed. The implementing steps includes: finding the maximum output gray scale value which can be expressed by the gray scale of each stage of equipment based on the image reproducing method of the output equipment (such as the size, shape, the degree of deep or light, arrangement of the output pixel and the grade of gray scale of the equipment), recording the output information of the image pixel in the corresponding interval of the gray scale equipment based on the probability expression of these two gray scales of the equipment from the output gray scale in the maximum interval of output gray scale corresponding to two neighboring gray scales of equipment, searching for the corresponding output information in the lookup table based on the result of transferring pixel color during outputting by the probability occupied in the two gray scales of equipment corresponding to the beginning and end of the interval of the output gray scale value of the output information record, whereby a better reproducing method for reproducing the pixel by using the output equipment can be determined.

Description

_ ^ £ 6712 ___ V. Description of the Invention ⑴ '----- [Application Field of the Invention] The present invention relates to a method for processing a rich image, particularly, a method for displaying a digital image, In the fields of printing or printing, it is a technique for optimal halftoning the pixels (P 丨 xe 丨) of an image. [Related technology of the invention] The half-color s-period (h a I f t ο n e) method is a technology widely used in the fields of image display, printing, and printing or publishing. Its advantage is that it can reproduce images with richer color levels in smaller device gray levels. For example, to print a color image on a black and white printer, and display different gray levels on a monochrome two gray level display. Degree-level images and more. 2. Traditional halftone processing methods mainly include threshold method, mode method, and oscillation method. The following uses the black and white grayscale printer image printing as an example to briefly introduce the implementation principles of these methods. (1) Threshold method:-Set a threshold (Threshold). If the true gray level of any pixel in the image is greater than this value, print it; otherwise, do not print it. The images produced in this way are only suitable for very few situations. (2) Mode method: Also known as the light-dotted dot map method, which uses a matrix to represent the gray level of a pixel 'for example, a 2 X 2 matrix can be printed by 0 dots, 1 dot, 2 dots' 3 Points or 4 points to represent 5 gray levels. In this way, when a grayscale value of one pixel is obtained, a matrix representing the grayscale is used instead. Similarly, if a matrix of 1 6 氺 16 can be used to represent 2 5 7 gray levels, this formula

Page 5 4 367 1 2 V. Explanation of the invention (2) The biggest disadvantage of the method is that it will enlarge the size of the original image. (3) Dithering: The oscillation method is divided into two types: pattern oscillation method and filter oscillation method. The pattern oscillation method is implemented by an oscillation matrix. The simplest method is to use an 8 × 8 matrix of 0,1. In the matrix, different gray scale matrices are obtained according to different distributions of 0 and 1. This principle is the same as described in the model method. Different from the mode method, the oscillating method does not correspond one pixel to one matrix, but corresponds one pixel color on the screen to one oscillating matrix representing the gray scale of the color, and uses the corresponding position element in the oscillating matrix. The value replaces the pixel to decide whether to dot or not (ie 0 or 1). Another method of the pattern oscillation method is the Bayer oscillation method, which is different from the simplest implementation method in that its pattern matrix has only an 8 X 8 matrix, and the elements filled in the matrix are numeric values 0 ~ 6 3 instead of 0, 1, This matrix can achieve 64 gray levels. In the original image, the pixels whose grayscale value is greater than the grayscale level of the corresponding position element in the pattern matrix are white pixels in the corresponding output image. Conversely, those pixels that are not greater than the pattern matrix Images are represented as black pixels. The most obvious disadvantage of the pattern oscillation method is that it is greatly affected by human factors and the effect is not good. In order to overcome human factors as much as possible, the most effective method is error diffusion. Consider a known pixel, and there is a certain error between it and the final representation result. If this error is spread to the surrounding pixels, its influence will not be too significant in the final output image. There are many potential problems with this method, especially the problem of finding the actual error of scattered pixels. Another more troublesome problem is to make

Page 6 V. Explanation of the invention (3) The original gray (¾-degree disparity area borders become soft and scattered. For example, an image that includes a certain essay is prone to blur at the edges of the text. S The most popular error dispersion The algorithm is F 1 0 yd-Steinberg filtering algorithm, and its working principle can be expressed as: X 7 3 5 1 The sum of all its numbers is 16. Set a threshold error between the pixel and black and white at X, Then you can add 7/16 of the error directly to the pixel located on the right side. (Add the 3/16 of the error to the pixel on the next row to the left of X, and 5/16 of the error to the pixel directly below X. Add the remaining 1/1 16 of the error to the pixels on the next row to the right of X. The halftone processing implemented by the oscillation principle does not require the original image to be enlarged. However, whether it is the pattern oscillation method or the filter oscillation method, The generation of their pattern matrix or filter is deeply affected by human factors and still cannot produce convincing better results. [Objective and Summary of the Invention] The main purpose of the present invention is to provide an By more The method of adding a smooth and delicate image to the halftone can also be free from the influence of man-made: 'u The technical means disclosed in the present invention, the implementation steps mainly include ·' 1 · According to the diagram of the output device Image reproduction methods (such as any one or more of the output pixel size, shape, density, arrangement, and device gray scale weaving number) and mathematical expectation formulas are obtained in any phase

Page 7 43 67 1 2 V. Description of the invention (4) The gray scale of the device varies with the output probability under the condition that the total output probability is equal to 1. (The output probability of the gray scale of smaller devices varies from 1 to 0, which is relatively Large device grayscale output probability changes from 0 to 1), the desired grayscale curve obtained, combined to get the entire output grayscale interval curve, which is a monotonically increasing curve; 2. the entire output grayscale interval curve Obtain output representation messages corresponding to each output gray level and level, and establish a comparison table to record the output representation information of pixels in the image at any gray level value; 3 * Obtain a pixel in the original image, From the grayscale value of the pixel, a corresponding output is found from the lookup table to indicate that the message 'half-tones this pixel according to the message until all pixels have been processed. In the technology disclosed in the present invention, statistical probability and expected value are used. First, the maximum output gray level that can be represented by the gray level of each level of equipment is obtained from the image reproduction (R epr 〇 duce) mode information of the output device, and the performance interval of the gray level of the device is divided into n non-overlapping regions. (Assuming that the device itself has η-level device grayscale), and the output grayscale value of the corresponding interval is represented by the grayscale of the two devices corresponding to the beginning and end of the interval according to the probability occupied. Introduce the expected gray level corresponding to each probability (that is, the most likely rotation gray level). Similarly, the most reasonable expression method of the expected gray level is to use these two equipment gray levels to express this distribution probability. We call the distribution probability of the gray scale and output probability of the two devices in the gray scale of the two devices as the output representation information of the output gray scale. Based on this principle, we can find the output representation information corresponding to each output gray level.

Page 8 3 G7 1 V. Description of the invention ¢ 5) The required output gray level establishes a comparison table for output information. At the time of output, a look-up table based on the gray level obtained by converting the color of the pixel is used to obtain the corresponding value of the pixel at the time of output. The output represents the message, and the pixel is represented in a better way. The detailed technical content and embodiments of the present invention will be described below in conjunction with the drawings. The aforementioned maximum output gray level can actually be determined in many ways, for example: a color pixel that contains three color values of R (Red), G (Green), and B (Blue). Its value range is 0-256, and a maximum output grayscale is 2 5 6 0 0 (the output grayscale interval is between 0 ~ 2 5 6 0 0) by converting in a way such as RX30 + GX59 + BX11. ). [Explanation of the drawings] FIG. 1 is a flowchart of main implementation steps of the present invention. Figure 2 shows that the printer can generate 13 uniform grayscale areas in a printable smallest square. Figures 3A to 3F are shown in the square in Figure 2. There are several possible combinations of printed dots. Fig. 4 is a flowchart showing the detailed steps of halftoning a certain color pixel. [Detailed description of the invention] Please refer to "Figure 1". The main implementation steps of the present invention include: 1. Obtain the gray scale level of the output device according to the image reproduction mode information of the output device. Under the condition that the total output probability of the gray level of the adjacent device is equal to 1, calculate the output gray level of the pixel at

Page 9 4 3 671 2 V. Description of invention ¢ 6)

: Corresponding area: 2, the machine occupied by the gray scale of the two devices corresponding to the tail;;, a desired output gray scale curve obtained by changing 卩 to lose, 〃 ', and combined to obtain a full round of gray Step interval curve; Secondly, according to the gray scale interval curve and line, a set of output indication information comparison table is established to record the pixels in the image at any output gray level value. And 3. Obtain a pixel in the original image, and output the grayscale value obtained by converting the color value of the pixel. Find the corresponding output representation message from the lookup table, and halftone the pixel according to the message until processing. Finish all the pixels. In the foregoing step 1, it is mainly based on any one or more of the image reproduction mode information of the output device (such as the size, shape, density, arrangement, and gray level of the output pixels). ) And mathematical expectation formula to obtain the gray level of any two adjacent devices under the condition that the total wheel output rate is equal to 1 'varies with the output probability (smaller device gray level output probability changes from 1 to 0' is relatively large (The grayscale output probability of the device varies from 0 to 1) to obtain the desired output grayscale curve, which is combined to obtain the entire output grayscale interval curve (this is a monotonically increasing curve). In the foregoing step 2, the so-called output representation information (halftoneinfo [rank]; rank represents the maximum output grayscale level of the output device), which includes: (1) an output grayscale value of a pixel in the image; (2) the gray scale levels of two neighboring devices for representing the output gray scale value; and

Page 10 V. Description of the invention (7) (3) The probability that the output grayscale value of the pixel is in the interval between the grayscale values of these two neighboring devices. " And the above steps (1) and (2) are executed in the initialization process of the out-of-equipment device to create a meta-indicating table of output information, which is stored in a readable memory device (such as memory or Other: similar components), so that the wheel-out device can use it when outputting halftone images. In the process of halftoning an image according to the foregoing method steps, the foregoing step (3) will be continuously repeated to obtain a pixel in the original image, which is obtained by converting the color value of the pixel The output grayscale value of is found from the corresponding table to find the corresponding output representation information, and the output representation information of the pixel is recorded, until all pixels are processed, the operation of halftoning the image is completed. "Explanation of the embodiment" Hereinafter, a color image is output on a multi-gray-scale printer. The process of half-toning an image according to the method disclosed in the present invention is described below. (A) First, it is explained how to generate the output gray-scale interval curve in the aforementioned "step 1". Assume that the image reproduction mode (printing method) of a monochrome (black and white) printer with multiple gray levels is that the output pixels are circular, the output pixels are arranged in a square, and the gray level of the device is It is 16 levels, and it shows how to achieve the maximum output gray scale of 2 3600. Assume that the gray scales of the corresponding 16-level grayscales of the printer are ’, where τ 0 is all black and 1 is completely white, and the corresponding device grayscale levels are sequentially

Page 11 4367 12 V. Description of the invention (8) 0,1,2,3'5,6,7,8,9,10,11,12,13,14,15. As shown in "Figure 2", suppose a minimum printable square with a side length a = 1, and the printable pixel positions are the four corner points P1, P2, P3, and P4 of the square. Each print The diameter of the pixel is r (, according to the printing conditions of the four corner points, the square can be divided into areas an, a12, a13, a14, a21, a22, a23, a24, a31, a32, a33, a34, a4 13 uniform grayscale regions, where an = a12 two,, a31 = a32 = a33 = a34 'a4 = A4, and their grayscales are hn, h12, h13, h14, h21, h22' h23 , H24, h31, h32, h33 'h34' h4. Then the average gray scale of this square can be obtained by multiplying the area of the 13 areas by the gray scale of each area, that is: // = + + a13 / il3 + ten "21Λ2Ι + + ίϊ: 3% 3 + ίϊ: 4Λ24 + Ο) (2)" 31 ^ 1 + + meaning 3 ^ 33 + 仏 > 4 ~ > 4 + ~ 乂 defines a grayscale interception Function F (X)

When 0 < X < 1 when m considers that the printing probability is distributed between two non-zero device grayscale ratios and h2 (hph), the relationship between the printing probability and the mathematically expected grayscale. Suppose the distribution of probability on h and h2 is 1-m, m (0 < m < 1). We equivalent this step to the sum of the two print results: (1) the original gray level is hi

Page 12 5. Description of the invention (9) Printing; and (2) Printing when the grayscale of the device is m. Total probability of pixels (m — 1) Probability of pixels in interval h2 The first part of the print result hu = h) 2 = h13 = h1,1 = F (hj) D2 2 = h22 = h23 2 h24 = F (2h 丨) h31 '= h32' = h33 '= h34' = F (3h,) h4 'Two F (called) The printing result of the two parts can be obtained according to the mathematical expression. Let ^ m ^ μ ^ ^ ^ pi ^ be the probability of m. Derive the relationship between the printing probability and the number of digits. According to the relative position between the printing point and the pixel, the number of stations and pixels The average gray level is Ph4, and the probability is m4_; there are three types of Y, and the average gray level is called, the probability is ⑹ The average gray level is phn, and the other two are diagonal forms. Remember Pk, each The probability of this situation m2 = m2 (lm) 2; column p sentence fire 1¾ and other combinations have α = 4, and the average gray level is ph〗 〖, out of prime, m 1 = m (1-m) 3 ; Print 0 pixels, this combination has p +: ^

The average gray scale is ph0 (phO = 0), and the probability of occurrence m0 = (1, m) 4. As shown in "° 3A figure 'Figure 3F'", it is the knowledge of the aforementioned several printing situations. * Inner characters 0 'I 2' 3 '4 indicate that the corresponding area is printed u, i, 2, 3 , 4 times. According to the above-mentioned average grayscale calculation formula (1), J Μ can be obtained

Page 13 436712 V. Description of the invention ¢ 10) The average gray levels of several combinations are: ρΚ = 0 ph [= \-(A) + 2. Az + j + A ^ jFfh) pool, = + A + 义 列 和 一 ⑷ + 24 十? ph „= ph, = \-(2A2 + 3AjF (h) -i2A2 + 3A: jF (2h)-(Ai + AjF (3h) ph ^ = XAA.Ffhj-AA.FflhjAA.FOhhA.Ff ^ h) expect The output gray level E [H] is: E [H] = x m4 χ phA + m3 x C34 x phy + m2 x (piu ^ x 4 + /? /? 22 x 2y + m, x C {x ρ / ϊ! + m0 x Cj χ ρ / ϊ〇 (J) This is the relationship between the probability and the mathematical expectation. Let us assume that η is the ratio of the radius r of the printing pixels to the square side length a, ie π = r / a, the area of A 丨, A2, A3, Ai [can be obtained., / «', rr jrr + ahlx Ύ is called Α Α2 = (^) 2 arctg ^ l-γ arctg] j ^ 4-even, η, 2 4 A = 3Α1-2Α2 + / ζτ (¾2 Α4-\ ~ πχ ^ ——Α, -2Α-, -3Α,

Page 14

I. Description of the invention (11) The two printing results are added to obtain ph〇 '' ph! ', Ph21', ph22 ', ph3', and ph4 ', respectively: ph' = l- [3A] F (hl ) + 2A2F (2hl) + A: F (3hi) + 3AiF (2h) + h7) + 2A2 F (\ + KJ + A, F (h2) + A, F〇h \ + h2)] ph2x '= \ -[2AxF (h () + A1F (2hx)-\-2A-1F (h ^ + h2) + 2A, F (2hxjrh,) + 2A, Ffh, + lh2) + A2F (2h2) + 2 ^ (^ ) + A, F (2h ^ IK)] ph22 '= \-[2AxF (h () + 2A, F (h2) ^^ A2F (hl + h1 _) ^ 2AiF (2hl + hJ) + 2A, F (hl + 2h2) + A4F (2h} + 2h2) J ph: = \-[AlF (hJ + 3A] F (h2j + 2A2F (h) + h2) + 2A2F (2h2) +

A, F (3h2) + 2A: F (2hz + h \) + A, F (2h \ + hj + A, F (h) + 3h2)] ph; = l-f4AiF (h2) + 4A2F (2h2) + 4A, F (3h2) + AiF (4h7) J Substitute the eight formula (3) to get the relationship between the expected output gray level and the probability distribution when the gray level interval distribution of the two devices is distributed: ~ x m4 x ph ^ '+ w3x C34 x ph,' + πιΊ y. (ph ^ 'y-A + ph, ^!) + m ^ C \ y.phx' + m〇x C ° Axp \ '(4) (B) Continue to explain how to establish in the aforementioned "step 2"-a kind of comparison table used to represent the information of the pixels in the recorded image at any output grayscale value.

Page 15 436712 V. Description of the invention (ί2) The rotation-out indication message (halft〇neinf〇 [rank]; rank indicates the maximum output gray level of the output device) according to the present invention, including: (1 ) The output grayscale value of a pixel in the image; we can use the index of the array. For example, an element halftoneinf〇 [l〇〇] in the array halftoneinfo has a subscript of 100, which means that it is the 100th element in the array; and in the processing method of the present invention, the gray scale is output. The output of a pixel with a value of 100 means that the message is stored in this element. (2) The gray level of two neighboring devices used to represent the output gray level value; if the smaller device gray level value is known, the larger device gray level value can be determined by the smaller device gray level Degree values are increased by 1 because they are adjacent and therefore can be represented by smaller device grayscale values. And (3) the probability that the output grayscale value of the pixel is within the grayscale value interval of the two adjacent devices; because the sum of the output probability of the grayscale value in these two devices is 1, The rotation probability of the degree value can be used to determine the rotation probability of another gray level value. It can also be expressed by the output probability of the smaller equipment gray level value. In the case of 5 texts, we use the whole /, do not have a chance. Therefore, the following virtual code example can be used to establish their data structure as follows: struct {int〇UtValuel; // Small output device pixel value ^ ㈣The output probability of the output pixel value of the device can be color if the original image is a color image Pixel of image

5. Description of the invention on page 16 (13)-A color value (R, G, B) is converted to an output grayscale value of level 2 5 6 0 0 by the following formula: Η-RX 30 + GX 59 + Β X 11; In this example, R (Red), G (Green), and B (Blue) are the components of three colors: red, green, and blue. The value ranges from 0 to 2 5 6. Therefore, the range of Η is L25600. The iterative method to create an output representation message comparison table h a 1 f t ο n e i n f 〇 [2 5 6 0 0] can be shown in the following example:

Void FillHalftoneInfoStruct (float η) {int hi = 0, h2 = 0, nCount = 0; for (int i = 0; i <15; i ++) {for (float j = 0; j <1; j + = 0.00001) ih 1 = (int) (CaculateExpertOutValue (η, i / 15, (i + l) / 15, j) * 25600); if (hi > nCount & & h2 < = nCount) {halftoneinfo [ nCountj.outvaluel = i; halftoneinfo [nCount] .outchancel = j * 25600; nCount ++;} h2 = hi;}

Page 17 4367 1 2 V. Description of the invention (14) Where the function Void FillHaiftonelnfoStruct (float η) indicates the establishment of an output representation message comparison table, η is the radius r of the printing pixels, and the distance between two adjacent printing pixels a The ratio, that is, n = r / a, (Vi < ”< 2), float Cacu1ateExpertOutVa 1 ue (float n, float hi, float h2, float chance) is used to calculate the expected output grayscale value, which is 〇 For floating-point values between ~ 1, chance is the output probability of the smaller device grayscale, and its calculation formula is as the aforementioned formula (4). Although this step is more complicated, but for a print output device It only needs to be initialized once, and the availability is very good. When η is determined, this formula is a unary quartic equation about m, and theoretically an accurate solution can be obtained. (C) About the previous step (3) Detailed embodiment: When the above-mentioned step (3) is actually performed for halftoning an image, the method steps disclosed in the present invention can be used to process each pixel of the image separately to obtain a better halftone. Toning results For example: A color pixel 'composed of three colors of R (Red), G (Green), and B (Blue)' can be halftoned according to the following steps (see "Figure 4"), The steps include: 1. Converting the color of a pixel to the corresponding grayscale value; assuming an output grayscale of 2 560 0 level is taken as an example ', it can be calculated according to the formula [rank = R χ 30 + GX59 + BX11; rank (0 < = rank < 25600)

Page 18 V. Description of the invention (15) The grayscale value corresponding to the color; 2 · According to the grayscale value obtained by the conversion, look up the corresponding rotation display message ha 1 ft 〇neinf 〇 from the output representation message comparison table. [rank]; Suppose that the 4 'turn-out representation message halftoneinfo [rank] = {6,1 5 0 〇0 丨, that means we can represent this one in one of the device gray levels 6 and 7. Pixel output gray scale, where the probability represented by the device gray level 6 is 1 50 0 0/2 5 6 0 0 (the probability is between 0 and 1), and the probability represented by the device gray level 7 1- (15000/25600) (probability between 0 ~ 1); 3 · Generate a random number r, such as using a random number generator (or function) to generate a random number between 0 and the maximum output gray level Random value Γ; (In this embodiment, it is to generate a random value between 0 and 2 5 6 0 0); 4 When the random value is less than or equal to the gray scale level of the device that outputs the message When the probability of the smaller one is occupied, the result of halftoning is the smaller of the gray scale of the device that outputs the message, otherwise The results for the adjustment of the output message indicates the gray level of the device plus a smaller, i.e. larger person. The above steps can be implemented by the following virtual code: int CovertToHalftoneValue (int R, int G, int B) {int rank = R * 30 + G * 59 + B * 11; if (Rand () < = Halftonelnfo [rank] .outchancel) return Halftonelnfo [rank] .outvaluel; else return HalftoneInf〇 [rank] .outvaluel + 1;

Page 19

Claims (1)

6. Scope of Patent Application1. An image halftoning processing method for halftoning pixels (pi X e 1) of an image, including the following steps: According to the image reproduction method of the output device Message 1 obtains the device gray level of the output device. Under the condition that the total output probability of the gray level of the two adjacent devices is equal to 1, calculate the desired output gray level as the output probability distribution changes. Output the grayscale curve, and then combine to obtain a total output grayscale interval curve; based on the output grayscale interval curve, establish an output representation message comparison table 1 to record the output representation information of all output grayscale values; And a pixel in the original image is obtained, and the output grayscale value obtained by converting the color value of the pixel finds the corresponding output representation message from the lookup table, and the pixel is represented according to the message, and this step is repeated until the processing is completed All pixels. 2 · The processing method for halftone of an image as described in item 1 of the scope of patent application, wherein the reproduction mode information includes at least the size, shape, density, arrangement, and gray scale of the output pixels of the output device Any one or more of the series. 3. The method for processing halftone of an image as described in item 1 of the scope of patent application, wherein the output indication message includes: an output grayscale; two adjacent ones of the output grayscale Device grayscale; and the output grayscale when printing is respectively provided by the grayscale output tables of these two neighboring devices P. 21 436712. 1 The processing method for image halftoning as described in item 3 of the scope of the patent application, ^ the printing probability of the smaller device gray level in the output gray level interval and the output gray level interval The sum of the printing probabilities of the parent gray scale of the device gray scale is 1. The processing method of halftone of the image as described in claim 1 of the patent scope, ', Medium. The Hai pixel system is a color pixel that contains three color component values of R (Red), G (Green), and B (Blue). The value range is 0_256. According to the processing method of image halftoning as described in item 5 of the scope of patent application, wherein the output grayscale value of the color pixel Η can be obtained by converting: H = R X 30 + G X 59 + Β X 11. The processing method of image halftoning as described in item 1 of the scope of the patent application, wherein the step of representing the pixel according to the output representation message further includes: converting the color of the pixel to the corresponding grayscale value; according to the conversion result Find the corresponding output representation message from the output representation message comparison table and generate a random number; the result of determining the halftone is the smaller of the gray scale of the output representation message equipment, when the random value is less than or When the output represents the probability of the smaller gray scale of the equipment representing the message; and the result of determining halftone is that the smaller gray scale of the equipment representing the output represents the message plus 1, when the random value Greater than the probability that the smaller of the gray scale levels of the device representing the message is occupied. Page 22 Page 23