TW201032578A - Method for enhancing image contrast - Google Patents

Abstract

A method for enhancing contrast of an image is provided herein. The method includes the following steps of: counting a first number of pixels of each gray level in the image; obtaining a number of none zero gray levels according to the first numbers; calculating a first threshold value according to the number of none zero gray levels and information of the image; converting the first numbers to second numbers according to the first threshold value; performing a Luminance Histogram Equalization (LHE) process of the image according to the second numbers so as to obtain a first gamma curve; obtaining a color characteristic value of each color of the image; tuning the first gamma curve to a second gamma curve according to the minimum of the color characteristic values; and matching the RGB color signals of the image with the second gamma curve so as to acquire a new image.

Description

201032578 P080365SSZ1TW 28724twf.doc/n VI. Description of the Invention: [Technical Field] The present invention relates to a method for improving image contrast, and in particular to giving appropriate contrast according to the characteristics of each image. Enhanced ways to increase image contrast. [Prior Art] Image processing is to color image information.

Darkness and other changes in various images. As people's requirements for digital image quality become higher and higher, the progress of digital image processing technology has also grown considerably. Regardless of the image that is too dark or too bright, the contrast (Contrast), brightness (Brightness), and color saturation (Saturation) of the digital image can be improved by appropriate image processing. A common way to improve image contrast is to use histogram equalization, where histogram equalization technique = enhanced image contrast and image brightness. Simple, say, histogram homogenization is to redistribute the bright f-values of all the elements of the entire image, so that the brightness and contrast of the entire image are more evenly distributed. The ^' histogram equalization technique can make the originally darker or brighter image = to a more normal color gradation. For example, if the image is too dark, more of the bright parts can be obtained by using the histogram, and the dark part can be obtained after the equalization. I used 10,000. However, the sentence homogenization technique is used to deal with the case where the energy is too concentrated, and the gray-scale discontinuity occurs. After the image is processed by the histogram homogenization technique, the color information of the image 3 201032578 P〇80365SSZlTW28724twf.doc/n color, saturation, etc. will be greatly different from the original image, and the histogram will be evenly distributed. The resulting image is severely distorted. SUMMARY OF THE INVENTION The present invention provides a method for improving the contrast of an image. The hunting value is used to adjust the total number of pixels corresponding to the gray scale in the process of homogenizing the histogram, so as to avoid the occurrence of gray scale discontinuity. Improve the contrast of the shadow ^. The invention further provides a method for improving image contrast, which is used to adjust the intensity of the histogram homogenization method, avoids the occurrence of gray scale discontinuity, and improves the contrast of the image. Furthermore, the present invention further provides a method of image processing by increasing the color saturation of an image to enhance the contrast of the image while making the image more vivid. In order to achieve the above and other objects, the present invention provides a method for improving image contrast, the method comprising: according to the brightness information of the image, the total number of first pixels Nj of each gray level in the image of the 10 leaves, wherein Nj is the gray P in the image. The total number of pixels whose value is specific to j, j is an integer not greater than L; and according to the counted total number of first pixels Nj of each gray level, the image - non-zero statistic gray scale number q · ' According to the non-zero statistic gray scale number q and the image information, a threshold value s is calculated; then, according to the threshold value s and the total number of first pixels of each gray scale Ν, a gray scale is obtained. The total number of two pixels Ni 'where i is an integer not greater than L, when Nj > S, Ni = S ' and when NjSS, NfNj; according to the obtained total number of pixels of each gray level Ni 'image A luminance histogram homogenizes the 201032578 P080365SSZ1TW 28724twf.doc/n (Luminance Histogram Equalization, LHE) program to obtain an image with a gamma curve; then determines a gray scale based on the total number of pixels p of the image and the predetermined number of gray levels L Distribution mean ul; The total number of sub-pixels of the parent-color sub-pixel in the gray image is Nj(c), wherein the variable c represents the color corresponding to the sub-pixel, and Nj(c) represents that the color corresponding to the sub-image is C. , the grayscale value is equal to the total number of sub-pixels of j; according to the external parameter St, the pixel slit p of the 彡 image, the average value U of the gray scale distribution, and the sub-pixel of each color in the total number of sub-pixels of each gray scale, The color feature values corresponding to each color are selected; and the gamma curve is adjusted according to the minimum value among the calculated color feature values. The present invention proposes another method for improving the ratio of the shadow ride, the steps of which include: the total number of first pixels per gray scale in the statistical image, where N is the total number of pixels in the image whose value is equal to j, 』 is not greater than [^^; according to the critical value s and the total number of pixels of each gray level n = per: gray total number of second pixels Ni, where i is an integer not greater than l 'Nj> S when 'N production S And when N# S, N two N.; the total number of second pixels per gray-scale obtained, and the image is like a two-dimensional histogram feathering material to enhance the contrast of the image. In addition, another method for improving the image contrast method of the present invention is based on the total number of pixels p of the image and the predetermined number of gray levels l; the average value of the two gray ash cloths; the number of times each color of the statistical image (8) 'Where the variable C indicates the total number of sub-pixels when the color corresponding to the j-th pixel of the sub-pixel is c, and j is an integer not greater than £. The receiver depends on the external parameter st, the total number of pixels of the image p, gray scale 5 SSZlTW28724twf.doc/n 201032578 jl ill and the sub-pixel of each color in the total number of sub-pixels of each gray level, calculate the color feature value corresponding to each color; and according to the calculated color characteristic values Minimum value, adjust the gamma curve of the image. In the present invention, the method further comprises: obtaining a non-zero generalization of the image according to the counted total number of pixels Nj per gray scale: a gray level q; and according to a non-zero statistic * gray The order (1 and the image of the image) counts out the critical value s. In the method of the present invention, the method of extracting the image contrast is as follows: its t non-zero breaking amount (four) number q # The gray _L subtracts the total number of gray scales corresponding to the total number of first pixels Nj in the image. In one embodiment of the present invention, the method for improving image contrast, wherein the image information includes image resolution and predetermined Gray scale number L. In one embodiment of the invention, the above method for increasing image contrast 'where the critical value s is equal to ux i+i , and u=£ L q_l 叩 q

In an embodiment of the invention, the above image contrast is improved

Method, ul=IL In an embodiment of the present invention, the method for improving image contrast, wherein calculating a color feature value corresponding to each color comprises:, according to a grayscale distribution average U1 and each color in the image The sub-pixel is calculated by the total number of sub-pixels of each gray level, and the overflow value corresponding to each color is calculated according to the external parameter St, the total number of pixels of the image p, the gray-scale distribution average ul, and the overflow value corresponding to each color. Calculate the corresponding color eigenvalues for each color 6 201032578 P080365SSZ1TW 28724twf.doc/n. In the invention of the present invention, the above method for improving the contrast of the image, wherein the overflow value corresponding to the color C is equal to | fine, when 顺), -G, and when Nj(6)>ul, △ mail) Equal to (10)) —ul). In the present invention, the b-, gay-containing method for improving the contrast of the image, wherein the color characteristic value corresponding to the color C is equal to

St v P-ul

I q(c)xst) L , and q(c)= 2^ANj(c). j=〇 method in two: three colors of the color of the image contrast enhancement ^ there is a knife corresponding to the red, green and blue squares 'in the embodiment, the above t:=, rRGB color space conversion to enhance image contrast To the order (4) - the number of pixels is based on the gamma statistics after the image is converted to the YUV color space. In the hair style-implementation towel, the above method for improving image contrast, wherein the number of pixels per gray-scale is based on red color information r, green color information, and blue of the image in the B color space. A color information in the color information B is statistically derived. Therefore, the present invention provides a method for improving the contrast of an image, which combines the intensity of the histogram equalization method and the limitation of setting the color saturation to avoid the occurrence of gray-scale discontinuities to improve the contrast of the image 7 201032578 j. uov /^^^SSZlTW 28724twf.doc/n degrees. The above described features and advantages of the present invention will be more apparent from the following description. [Embodiment] FIG. 1 is a flowchart of a method for improving image contrast according to an embodiment of the present invention. Referring to Figure 1, first in step S11, an original image is taken. The original image has a total of pixels P (pixels) and is displayed with a predetermined gray level L·, where L is a positive integer. Next, in step S120, the original image is converted from the RGB color space to the γυν color space, where “γ” represents Luminance, and “U” and “V” represent Chrominance and chroma, respectively. (Chroma). Next, in step S130, a first histogram of the original image is established according to the total number of first pixels Nj of each gray level in the image of the "Y" information, wherein the pixel of the original image with a grayscale value equal to j is represented. The total number, j is an integer not greater than [. Next, Lu will use FIG. 2 to illustrate the first histogram of the original image. In Fig. 2, for convenience of explanation, it is assumed that the original image can display the gray level of the ith order to the first order, that is, the predetermined gray level L of the original image is 1G order. Of course, the predetermined gray scale number LJi is not limited thereto. In the non-realistic towel of the present invention, the predetermined gray scale number L may be other positive recordings, for example, the predetermined mirror number L may be 256, to represent 0 to a total of Gray scale of 256 steps. Further, in the present embodiment, the total number P of pixels of the original image is 100, in other words, the resolution is 1 像素 pixel. The total number of pixels corresponding to each gray level is as shown in FIG. 2, where Nj is the total number of pixels of the grayscale value equal to j, for example, the grayscale value is equal to 4, 201032,078, P080365SSZlTW28724tWf.doc/n, the first pixel The total number & equal to 8, the first pixel with a grayscale value equal to 7 is equal to 35. , 'The following description, please refer to the figure 丨 and Figure 2. In step 814, the total number of first pixels Nj of FIG. 2 is obtained to obtain a non-zero statistic gray scale number = non-zero statistic gray _ q is the predetermined gray · L minus image = the total number of pixels to be called is zero The total number of gray levels. Take this real country 2 as an example. The i-th gray scale of Fig. 2 corresponds to the first image of the 1G gray scale. The total ^:N々Ni〇 is zero, and the predetermined gray scale number is [1, so Ji Shi The gray order q is 10_2=8. Next, in step s1, 〇, = non-zero statistic gray scale q and the original image information, the estimated critical value κ is according to the present invention - the critical value of the equation 8 is expressed as follows: I '· - q S = ux 1 + ^ Medium / indicates _ oblique yell, zero statistical order. The above-mentioned non-zero statistic mean w U --- q total i in the 'q table 计量 计量 gray _, p 影像 image / = in 'in this embodiment' because the total number of pixels is (10), then non-zero The statistic average is in this embodiment, the predetermined gray scale number 1 is 1 〇 order, the average value u is 12.5 ′ instead of the zero statistic gray scale number q, then “order°, 201032578 r^oujojoSZlTW 28724twf.doc/n to S =12.5*[l+(10/8)卜28.125, after the four masks are entered, the critical value S is obtained as 28. In addition, the process of calculating the critical value s may not pass through the non-zero statistic mean u, directly by The non-zero statistic gray scale number q, the image resolution, and the predetermined gray scale value L are used to calculate the critical value s. Next, in step S16G, the second histogram is pushed according to the total number of pixels and the 6a boundary value S. The method is that if the total number of first pixels N" exceeds the critical value S, the total number of the first pixels is reduced to the critical value S', and if the total number of pixels corresponding to the gray level does not exceed the threshold value S 'The number of pixels remains the same. In other words, assuming that the total number of modified pixels is the total number of pixels of the second pixel Ni, then Nj>s, wind = !§, and when ^S, 'NrNj', the towel i is an integer not greater than L. Thereby, the second histogram is obtained according to the first histogram (Fig. 2) and the threshold S. For example, the threshold s calculated by the embodiment is 28, and the gray level value of the first histogram is equal to the total number of pixels of the seventh pixel n7=35, and the total number of the first pixels of the grayscale value branch 8 is more than 33. The critical value s, so the total number of pixels of the 7th and 8th orders is revised down to 28, while the total number of remaining pixels below the critical threshold s remains unchanged, and the modified second histogram is as follows. Figure 3 shows. Next, in step S170, a Luminance Histogram Equalization (LHE) program is performed on the original image according to the obtained total number of second pixels of each gray scale to improve the contrast of the image. In detail, in step s17, the second straight: graph will be homogenized to update the gamma information, and then the updated Y = 5fl. Finally, in step S180, the updated Y, the information and u, V are δί1 are converted to the new color information R, G', B', and a contrast is obtained 201032578. The brain is allowed to be 1TW28724twfd〇c/n. New image. Figure 4, (4) domain VO, Figure 4 _ original image of the square image of the original image is processed by the conventional histogram homogenization technique and Figure 4 (C) is the original image by the method of the present invention, compared with the figure, and Figure 4 (〇 can be found = parametric, xi strict, talk domain rhyme histogram, its gray with discontinuity: the histogram processed by the method of the present invention due to the use of 'so there is almost no gray-scale discontinuity. The step of establishing the first histogram of the original image is performed in redundancy & however, in other embodiments of the invention or in the red color/bea color information G and blue color information B of the image in the RGB color space The first part of the f: color information is obtained by the histogram. The detailed method of the statistical dance statistic gray number q can be shown and explained with reference to the following flow chart. ❹ (4) Please 5' Figure 5 is to obtain non-zero statistic gray scale The method of the number flow is private. In the example of H, the lowest gray level of the face is set to 0. As shown in Fig. 5, first of all, in the frequency S41G towel, when the material is zero, the gray number q is first The non-zero statistic gray scale Sq and the gray scale j return to zero. Next in step s43 If it is determined whether the total number of pixels Nj is equal to the number Nj is greater than zero, then step is performed to add non-zero_quantity=:q to 1'. If the total number of pixels β is not greater than zero, step S450 is performed to determine the grayscale number j. Whether it is equal to the predetermined gray scale number l, if the gray scale number J is not equal to the predetermined gray scale number L, then step S42 is performed to add the current gray scale j to the step S430 to S45, and then increase the gray scale one by one. The order i; on the other hand, if the gray level order i is equal to the predetermined gray level number [, 11 201032578 P080365SSZ1TW 2S724twf.doc/n, the end operation obtains a non-zero statistic gray scale number q. Second Embodiment FIG. 6 is a diagram according to the present invention. In another embodiment, a method for improving image contrast is shown in FIG. 6. First, in step S510, an original image is received, the image having a plurality of pixels of total P, and each pixel includes a corresponding color. a plurality of sub-pixels (sub_pixel), the above colors may be red R, green G and blue B, and each sub-pixel displays 'P and l as a positive integer with a predetermined gray level L·. Then in step S52〇 • In, will analyze the received image To obtain the image data on a screen between the RGB color space, and calculate the average gray level distribution equation ul wherein the gray level mean of the distribution can be expressed as follows ul:. Ul = ~

L where P represents the total number of pixels of the image and L represents the predetermined number of gray levels. In the example of the present embodiment, for the sake of convenience, it is assumed that the original image can display the gray scale of the 10th order of the i-th order t, that is, the predetermined gray scale number of the original image is 10 • order. Of course, the predetermined gray level number L· is not limited thereto. In addition, in this embodiment, the total number P of pixels of the original image is 100, that is, the resolution of the original image is 1 〇〇 pixel. Therefore, in this embodiment, the average value ul of the gray scale distribution is 100/10=10, and then the three color information of r, G, and B are respectively calculated, and the red color information R is first used as an example of the calculation process. After obtaining the red color=寊 R R, in step s53 ,, the red color information R is counted in the total number of sub-pixels Nj(r) of each gray scale to establish an R histogram in which the total number of sub-pixels Nj ( r) r represents red color information R, Nj 12 201032578 P080365SSZ1TW 28724twf.doc/n The total number of tables is not red color information R in the gray level value is equal to the sub-pixel of j 'following the 5th history of the R histogram 7 shows the total number of sub-pixels Nj(r) used to represent red, for example, N4(r)=6, N6(r)=28. Next, in a small step S531, a red overflow value q(r) is obtained from the R histogram and the grayscale distribution average value U1. The equation of the red overflow value q(r) can be expressed as: ΣΔΝ/γ) j=0 ❿ ❹ /, medium, g Nj(r)Sul, △N/r-o, and when Nj(r)>ul When ANjO) is equal to (Nj(r) - ul). Therefore, the red overflow value q(r)g R is the sum of the average ul(10) primes of the gray-scale distribution of the pixel remaining in each of the gray levels in the histogram. Taking Figure 7 as an example, the case where the average value of the gray scale distribution ul is equal to 1〇2, the towel has only the sixth-order gray scale, the seventh-order gray scale and the eighth-order gray scale, and the number exceeds the gray-scale distribution average. Ul, the values are π, and. Therefore, compared with the average value ul of the gray scale distribution, the first number is more than 18, the 7th order gray heart back center f P white and P white pixel gray p 比 ^ ^ ^ gray pixel number exceeds 20, the 8th order fire h The number of pixels exceeds 22, 18+2〇+22=60. The red overflow value q(r) is the leveling in step S532. According to the equations of the external parameter st, the gray level sub-U1 and the red overflow value coffee knife, the red characteristic value rsp can represent the following color characteristic value 邶rsp = P-ul Let, st denotes that the external parameter is expressed as the average value of the gray scale distribution, and p /Of:: the scratch color overflow value 'P is the total number of pixels of the image. 13 201032578 P080365SSZ1TW 28724twf.doc/n In this example, the external parameter (pair) is set to 4〇, so the red characteristic value ^叩 is 40-[(60*40)/(10〇-1〇)] = ΐ3·33 After rounding off, the red eigenvalue rsp is 13. Further, the steps S533 to S535 for calculating the green characteristic value gSp and the steps S536 to S538 for calculating the blue characteristic value bsp are similar to the steps S530 to S532 for calculating the red = value rsp. Wherein, as long as the green overflow value q(g) and the blue overflow value q(b) are substituted for the red overflow value q(r), the green 4-inch eigenvalue gsp and the blue eigenvalue bSp can be obtained, and the green eigenvalue is obtained. The equation for gSp and blue eigenvalues is expressed as follows: bsp = st-S^l P-ul gsp = St q{g)^st~P-ul where q(g) is expressed as a green overflow value, q(b) The equation expressed as a blue overflow value 'and the green overflow value q(g) and the blue overflow value q(b) can be expressed as:

q(g) = fJANj(g) j = 〇 q(b) = fjANj(b) j = 0 It is assumed that the green eigenvalue gsp and the blue eigenvalue bsp obtained after the above operation are 12 and 15, respectively. Next, in step S540, the red feature value rsp, the green feature value gSp, and the blue feature value bsp are compared, and the minimum value of the three t is the second threshold SP, so the second threshold SP of the embodiment is 12. Next, in step S550, according to the second critical 201032578, the horse curve is used to replace the qi and qi 珈 curve.

The value sp replaces the first imaginary curve of the image. In step S560, the gamma curve obtains a saturated and contrasting image. - In addition, FIG. 8 is a flow chart of a method for obtaining a red overflow value. In the embodiment, the lowest gray level value of the facet is set to zero. As shown in Fig. 8, when the red overflow value q(r) is started in step S710, the red overflow value q(r) and the gray level j are reset to zero. Next, in step s73, it is judged whether the total number of readings of the red color information R is greater than the grayscale distribution average ul; if the number of records_ton is greater than the average value of the grayscale distribution, then step S· is performed: The difference between the average value of the number of U1 is added and accumulated; otherwise, the total number of images is 1; if Nj(r) is not greater than ul, then step S75 is performed to determine whether the gray level is equal to the pre-ashing & number L, if grayscale If the order j is not equal to the number of reserved gray levels L, proceed to step S720 'add the current gray level j to the step S73 to S750, and increase the gray level one by one], and the total number of accumulated secondary images is called (7) and gray. The difference of the average value ul of the order distribution; conversely, if the illusion is equal to the predetermined gray level number L, the end operation obtains the red overflow value q(r). The calculation method of the green overflow value q(g) and the blue overflow value q(b) can be obtained by the above method, and will not be described here. THIRD EMBODIMENT Figure 9 is a flow chart of a method for enhancing image contrast in accordance with another embodiment of the present invention. This embodiment is to combine the brightness histogram homogenization of the first embodiment and the method of adjusting the color image saturation of the second embodiment to improve the contrast of the two images. Referring to FIG. 9, in the method of 201032578 P080365SSZ1TW 28724twf.doc/n, the method of obtaining the first critical value s and modifying the first histogram of the original image according to the 苐-threshold value s' to obtain the second The steps of the f diagram are taken from the first - the actual customs, and are described in detail. After obtaining the first histogram, then in step S870, the obtained second histogram is homogenized to update the gamma information of the original image, and the updated Υ' information is obtained, and according to Υ, the information is obtained first. Gamma curve. Further, in steps S821 to S851, the operation for obtaining the second critical value SP for adjusting the color saturation is the same as the second actual value, and will not be described herein. Therefore, after obtaining the first gamma curve and the second critical value ’, then proceeding to step S880, the first gamma curve is modified according to the second critical value sp to obtain the second gamma curve. Finally, in the step, the original RGB color information is obtained corresponding to the second gamma curve to obtain a new RG'b' color transmission, and a new image with higher contrast is obtained. In addition, according to the above embodiment, the first critical value s 盥 3 boundary can be summarized. Please refer to FIG. 10 and FIG. 11 below. Fig. 10 is a flow chart of the s-meter value of the squad-technical value. Please refer to the following and the following description. First, in step S61G, the RGB^ of the image is converted to the γυν color space to obtain the Y information therein. Then, in step S630, the total number of pixels of each gray scale is calculated. In step S630, according to the total number of pixels of each gray scale, the number of pixels is non-zero statistical gray number q. Next, in step S64G, the root J statistic gray scale number q and the total number of pixels of the image p are calculated as non-average values 『. Finally, in the _ towel, the root _ zero lining gray ^ =, the non-zero statistic mean and the predetermined gray value Lt + calculate the first - S. In addition, in the above step S63, after obtaining the non-zero statistic, it is also possible to go directly to step S650, according to the non-zero system! 16 SSZ1TW 28724twf.doc/n 201032578 The number q, the total number of pixels P of the image and the predetermined gray scale value L calculate the first critical value S.

Figure 11 is a flow chart for calculating the second threshold value. Please merge the formula of the reference map η with the following description. In step S931, the red color information R of the image is obtained. Next, in step S932, the total number of sub-pixels Nj(r) of the red color information R is counted. Further, in step S9l, the grayscale distribution mean ui is calculated based on the total number of pixels P of the image and the predetermined number of gray levels L. Next, in step S933, the red overflow value q(r) is calculated based on the total number of sub-pixels (7) and the gray-scale distribution average value ul. Next, in step S934, the red feature value rSp is calculated based on the external parameter st, the red overflow value q(r), the calculated gray scale distribution average ul, and the total number of pixels P § ten. Similarly to the above-described steps S931 to S934 for calculating the red characteristic value rsp, the green feature value gsp and the blue feature value bsp are obtained via the calculation steps milk 41 to S944 and steps S951 to 954. Finally, in step S960, the smallest of the three values is the second critical value SP compared to the red feature value rsp, the green feature value gsP, and the blue feature value bSp. Bu '"杈咼景景> like contrast method, according to the characteristics of each image, calculate different parameters (such as the first - threshold ^ second threshold) 'to (four) like non-degree Contrast enhancement to avoid the situation where the traditional histogram uniformizes the gray-scale discontinuity. In addition, adjustments may be made to the extent that the color of the image in the prior art has been lost as described above in the preferred embodiment. However, it is not intended to be used as the scope of protection of the present invention; = circumference 17 201032578 P0S0365SSZ1TW 28724twf. The definitions defined by doc/n shall prevail. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a flow chart showing a method of enhancing image contrast according to an embodiment of the present invention. 2 is a first histogram in accordance with an embodiment of the present invention. 3 is a second histogram in accordance with an embodiment of the present invention. Figure 4 (A) is a histogram of the original image. φ Figure 4(B) is a histogram of the original image processed by the conventional histogram homogenization technique. Figure 4 (C) is a histogram of the original image processed by the method of the present invention. Figure 5 is a flow chart of a method for obtaining a non-zero statistic gray scale number. 6 is a flow chart of a method of enhancing image contrast in accordance with another embodiment of the present invention. Figure 7 is a histogram of red color information r in accordance with an embodiment of the present invention. Figure 8 is a flow chart showing the method of obtaining the non-zero order of the red color information R. ® Figure 9 is a flow chart of a method for enhancing image contrast in accordance with another embodiment of the present invention. FIG. 10 is a flow chart showing a first threshold value calculation according to an embodiment of the invention. Figure u is a flow chart of a second threshold calculation according to another embodiment of the present invention. [Description of Main Component Symbols] S110 to S180, S410 to S450, S510 to S560, S610 to S650, S710 to S750, S110 to S890, and S910 to S970: Step 18

Claims (1)

201032578 ruou^ojSSZlTW 28724twf.doc/n VII. Patent application scope: L—a method for improving the contrast of an image. There are a plurality of pixels with a total of P, and each pixel contains multiple pairs of different colors. Sub-pixel (Sub_pixel), each sub-pixel is displayed with - order L·, P and L are positive integers, and the method includes: ## according to a brightness information of the image, counting the first of each gray level in the image - The total number of pixels Nj 'where Nj is the total number of pixels in the image whose grayscale value is equal to 』 is an integer not greater than L; 0 according to the total number of first pixels of each grayscale, the image is obtained. The non-zero statistic gray scale number q ; 'According to the non-zero statistic gray scale number q and the information of the image, a critical value S is calculated; according to the critical value S and the total number of first pixels of each gray scale Nj, A total number of second pixels Ni of each gray level, where 丨 is an integer less than L, when Nj>S, NfS, and when N# S, NrNj; according to each gray level obtained The second pixel total wind, the redundancy histogram uniformity of the shadow image (Luminance Histogram Equalization, LHE) program; determining a grayscale distribution average value U1 according to the total number of pixels P of the image and the predetermined gray scale number L; and counting the subpixels of each color in the image in each grayscale The total number of sub-pixels Nj(c) 'where the variable ^ represents the color corresponding to the sub-pixel, and Nj(c) represents the total number of sub-pixels whose gray-scale value is equal to j when the color corresponding to the sub-pixel is c; The parameter st, the total number of pixels of the image P, the gray level is 19 201032578 ruou^ojoSZlTW 28724twf.doc/n the cloth average ul and the sub-pixel of each color in the total number of sub-pixels of each gray level 'calculate each color Corresponding color feature values; and adjusting the diagnostic gamma curve according to the minimum value among the calculated color feature values. 2. The method of claim 2, wherein the non-zero statistic grayscale number q is equal to the predetermined grayscale number L minus the image; the first pixel total number Nj of the first pixel is zero grayscale total. 1. The method of claim 1, wherein the image comprises: (4) the resolution of the image and the predetermined gray level B. 4. The method of claim 1, wherein the critical value S is equal to ux 1 + i, and "!: L q" q 5. The method of claim i, wherein. The method of claim 1, wherein the leaf calculates a color feature value corresponding to each color, including: ef 10 ^ according to the gray scale distribution average ul and each color in the image For each gray-scale total number of sub-pixels, calculate the overflow value corresponding to each color; ^ according to the external parameter poor, the total number of pixels of the image p, the gray-scale average value ul and the overflow value corresponding to each color , the color characteristic value corresponding to the color. ^ 7. The method according to claim 6, wherein the color c corresponds to an overflow value equal to, when Nj_ul, △ just two 20 201032578 ruoujojSSZITW 28724twf.doc/n 〇, And when Nj(c)>ul, ANjCc) is equal to (Nj(c)-ul). 8. The method of claim 7, wherein the color cf + q(c)xst, corresponding The color eigenvalue is equal to, and q(8) = i > Nj(c). j=0 9. The method of claim 1, wherein each pixel contains a corresponding color corresponding to red, green, and blue. Three sub-pixels. 10. - used to enhance the contrast of an image The image has a plurality of pixels and is displayed by a predetermined gray scale number L, where L is a positive integer. The method includes: counting a total number N of first pixels of each gray level in the image, where N is gray in the image The total number of pixels whose order value is equal to j, j is a small integer; according to a threshold value S and the total number of first pixels Nj of each gray level, • a total number of second pixels of each gray level is obtained, where i For an integer not greater than l', when Nj>S, Ni=S, and when NjSS, Ni=Nj; and according to the obtained total number of second pixels of each gray level, a luminance histogram is applied to the image. A Luminance Histogram Equalization (LHE) program for improving the contrast of the image. 11. The method of claim 10, wherein the information of the image includes the resolution of the image and the predetermined gray level L. 12. The method of claim 1, further comprising: converting the image from an RGB color space to a YXJV color space; 21 SSZlTW28724twf.doc/n 201032578 wherein the first pixel of each gray level The number of winds is based on 13. The method of claim Y, wherein the first pixel number of each gray scale is based on the red color of the image in the RGB color space. The color information of the information R, the green color information G and the blue color information B is statistically obtained. 14. The method of claim 10, further comprising: according to the statistics of each gray scale a total number of pixels Nj, find this a non-zero statistic gray scale number q of the image; and calculating the critical value S based on the non-zero statistic gray scale number q and the information of the image. 15. The method of claim 14, wherein the non-zero statistic grayscale number q is equal to the predetermined grayscale number L minus the total number of grayscales corresponding to the total number of first pixels Nj in the image being zero. 16. The method of claim 15, wherein the application The boundary value S is equal to ux 1 + L q where u = |, p is the total number of HL pixels of the image. 0 17. A method for increasing the contrast of an image having a total number of P pixels, and Each pixel includes a plurality of sub-pixels corresponding to different colors, each sub-pixel is displayed with a predetermined gray level L, and P and L are positive integers. The method includes: according to the total number of pixels of the image P And the predetermined gray scale number L, calculate a gray scale distribution average ul; 22 201032578 ^SZlTW28724twf.d〇c/„ (4) The image sub-pixel of each color is in the sub-pixel of each gray scale, wherein the variable C represents The color corresponding to the pixel ^ When the color corresponding to the second pixel is C, the grayscale value is equal to the total number of the under-pixels of j, and j is an integer not greater than L; the basis of the external parameter st, the image of the image = mean ^ and the color eigenvalues of the sub-pixels of each - color in the Si % number of each - gray scale 4 - different output - color; and - according to the calculated material curve of each color lag value #. 18. The method of claim 17, wherein ul:f. 19. The method of claim 17, wherein the color characteristic values corresponding to each color include: ^ according to the gray level distribution average ul and the image A-color sub-pixel in each gray_ Sub-pixel stitching, calculating the overflow value corresponding to the gorge; according to the external parameter st, the total number of pixels of the image p, the average gray value of the gray-scale distribution ul, and the overflow value corresponding to each color, the corresponding two colors are calculated. Color eigenvalue. 2 〇 · The method of claim 19, wherein the color c corresponds to an overflow value equal to | Ke, when Nj (la ul, = 〇, and when Nj (c) > When ul, ΔΝ"(6) is equal to (Nj(c) - ul). 21. The method of claim 20, wherein the color 23 201032578 a v^JSZlTW28724twf.doc/nr ^ q(c)xst^c The corresponding color feature value is equal to, for _p_ui^, and q(c)=^> ming. j=〇22. The method of claim 17, wherein each pixel contains a difference Corresponds to three sub-pixels of red, green, and blue. twenty four