TWI274908B - A method for interpolating the color of color filter array - Google Patents

Abstract

A color interpolation method for color filter array (CFA) is disclosed. After sampling the image through the color filter array, this method introduces pre-estimation the minimum square error (MSE) to address the color interpolation for color filter array. In order to estimate the missing pixels in color filter array pattern, the weights of adjacent color pattern pairs are decided by the matrix computation. In the operating, this method also adopts the one color model to achieve better human visual effect.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention This invention relates to a method of image color interpolation, and more particularly to interpolating color 利用 using a pre-estimated least square error (Mean Squafe Muggle (10); MSE). A method of color interpolation of an image obtained by an array of optics. Φ [Prior Art] In recent years, digital cameras (Digitalst out cameras; DSCS) have become very popular consumer electronics products. Digital cameras generally use a Charge C〇Uple Device (CCD) or a complementary MOS semiconductor: (CMOS) sensor to convert the sensed incident light into an electrical signal. The analog/digital conversion becomes a digital image signal. In order to save the size of the sampled color pixels and sensors, it is common to use a color filter array (Color Filter Array; (10)) as a type: _-like format, so a color filter array is coated on the CCD or Cm〇s on the side sensor. The commonly used color filter arrays are Bell (Βπα £. The CFA published by (4) is shown in Figure i. # 中中r is red, g is green, and B is blue. In CFA, each sampling point has only one color, that is, the full color system (ie, the three colors of red, green, and blue), and because human vision is sensitive to Luimnance or Green channel. In the whole [FA, half of the filters are green. Then, by appropriate processing, the basic pigment components contained in the pixels to be interpolated are processed by interpolation to 5 1274908. Real image results. - The conventional interpolation method refers to determining the unknown value of the function in some known sampling points. Traditional interpolation methods are very many, such as nearest neighbor (Nearest Neighbor) interpolation, double line Interpolation methods such as BUinear interpolation, cubic • B-SPHne (Cubic B-spHne) interpolation and cubic convolution (a- • C_ — ) interpolation, but these traditional interpolation methods have their own disadvantages, for example, Nearest point interpolation and bilinear interpolation Although the speed is fast, but the quality of the interpolation compensation is generally low, the image after interpolation often appears severely blurred, and there is a serious fatigue in the boundary image area (ie, the mineral tooth system is not good enough). Image quality, in the current video image is not high resolution, high contrast, the nearest neighbor interpolation and bilinear interpolation is not suitable. And three B-spline interpolation and three cyclotron interpolation The calculation of the method requires too many parameters for multiple operations, so the calculation process is too complicated. In the image signal processing system, silk A匕# Μ, Shishan, and Xiong, although providing accurate and good quality interpolation Participation, 'But because of the complexity and time-consuming operation, the cost of the product will increase when implemented' and it is not suitable for the instant digital color image acquisition system. Moreover, the digital color image acquisition system using the CCD and CFA sampling formats In the application, traditional interpolation method, the interpolated image is prone to color noise and edges, and blur, which is the main disadvantage of the traditional interpolation method. Therefore, it is necessary to provide a new type of Like color interpolation technology, the color of a color shoulder array in a digital image processing system (for example, a digital camera or a digital camera) can be interpolated into a higher quality image, and a rule algorithm is provided. It is advantageous for a digital signal processor or an integrated circuit. 6 1274908. [Invention] Therefore, in accordance with the above object, one of the objects of the present invention is to provide a method for image color interpolation, which is a color interpolation method. The color image of the Color Filter Array (CFA) is interpolated and restored using a pre-estimated least square error (MSE). A color model (KR, KB) is also used in the process to make the restored color closer to the human visual response. Therefore, the present invention has higher image quality than the prior art, and is more prominent in the effect of blue pixels. Another object of the present invention is to provide a method of image color interpolation which utilizes adjacent color pixel weights (Weights) as determined by matrix operations. Since the matrix operation of the present invention is relatively regular, it is suitable to implement the algorithm of the present invention on a hardware. . According to the above object of the present invention, an image color _ is included: Included: an array of chessboards having a plurality of odd-numbered columns and a plurality of even-numbered columns arranged in a staggered manner. One pixel, a plurality of first pixels, and a plurality of third pixels, one of the two pixels of the first pixel and the first pixel are arranged in an odd column, and the other part of the first pixel and the second pixel of the first pixel The system is arranged in an even column, where the second brother-pixel mother is horizontally and immediately adjacent to the first silly, each of the "or" or the second pixel. The mother has a first primary color, a first color, a second color interpolation value, and a second color interpolation value, and each of the first and second mothers has a second primary color value, a first 7 1274908 color interpolation value and a first - flute _ Λ βπ 4: ^ a color interpolated value, each third pixel has a third primary color value, a second color-interpolation value and a third second color interpolation; Using an interpolation step to target the first-target pixel of the first pixel, i Λ The first and third color interpolated values of the pixels of the first target are calculated by four pixels of the twelve third image masters and the younger brothers adjacent to the first target pixel horizontally or vertically adjacent to each other. Using the interpolation step 'to target the second target pixel of the second pixel, using the second pixel in the second - second, 'different private pixel's nearest twelve twelve images': 庶冬弟像▲ prime: four horizontally or vertically adjacent to the second target pixel: one, one: the second third color interpolated value of the second target pixel; the sound φ two one interpolation step 'for the first target a pixel, using the second pixel of the second image of the second/second::: the second pixel adjacent to the diagonal of the pixel, the first pixel 4a & ^ ^, the diagonal of the pixel of the first pixel and the pixel of the first target The first and third interpolated values of the five pixels of the position, and the four first pixels adjacent to the diagonal of the silly Quaidine in the first pixel, A standard like the quaint 豕f brother - the second color interpolation; the fourth interpolation boat uses the second pixel two: The third target pixel of the third pixel '.^ is the second pixel of the second target pixel, the first pixel of the first pixel is the most adjacent to the third interpolated value. The first one of the six first pixels of the target pixel is ^. ^ The second color interpolated value, the first third color interpolated value of the first pixel, and the pixel level of the third one. The first two-color interpolated value of the two adjacent first pixels, and the second second color interpolated value of the second main color of the second main pixel of the second main pixel are perpendicularly adjacent to the third target pixel. ^, the second third color interpolated value of the second pixel, and the third pixel in the third pixel horizontally and vertically as the third target pixel, to calculate the third target pixel - 2 of the second pixel Interpolating the first color; inserting the step by the fifth, with the second pixel of the first image: interpolating the value of the pixel of the second pixel and the pixel of the second pixel. The diagonal lines of the five second pixels of the pixel position of the pixel are adjacent to each other and are interpolated in the first image of the image. And the second ', ', and the next diagonal of the pixel diagonally the nearest w ^ $ to calculate the first - 庐μ# Guzhou heart four brothers two pixels, to count the younger-private pixels in the second - color Interpolating; and symplectic interpolation step, for the third target pixel, using the first pixel of the first-image::pixel, and the first two-color interpolated value of the second pixel being the most adjacent to the second target In the second pixel of the private-private pixel, the second-third-color interpolated value, and the second-third interpolated value of the second pixel are The third color interpolation, the first third color interpolating value of the first pixel, the third color interpolating value of the two first pixels of the vertical phase 4 of the third target pixel, and the horizontal sum of the third pixel The third first color interpolated value of the third target pixel is calculated by vertically adjacent to the fourth third pixel of the third target pixel. [Embodiment] The present invention relates to a method for image color interpolation, and in particular to interpolating an image obtained in a color filter array by using a Meail Square Minimum (MSE). A method of color interpolation. The operation of the preferred embodiment of the present invention will be discussed in detail below. However, the embodiments disclosed herein are not intended to be used and utilized in the present invention. The specific embodiments discussed are merely illustrative of the specific aspects of the invention, and are not intended to limit the scope of the invention or the scope of the appended claims. Referring to FIG. 1 again, the image color interpolation algorithm of the present invention is started by the following formula (1), wherein the ruler (1) is the actual value of the center point, especially, the value of the pixel is the center point, and κ (χ) is the actual value of the neighboring pixel ,, where i, · is the index value (Index) of the neighboring pixel, Wi, Wj is the weight value of the neighboring pixels, and η is the number of neighboring pixels. = (1) The minimum difference between the actual value of the center point ^ and the estimated value ^"(χ) is defined as (2): " k{xyK(x)j) η η η 1 7-1 /= 1 (2) η η η =Σ Σ wiwjc{x^ ^ 2Χ wtc(x, xt) + C(〇) )=1 ζ=1 square error, the step is derived from where ¢:(^7.)= ^(χ,)·尺(Ά, in order to get the minimum flat (2) partial differential as shown in equation (3) and let the result be equal to zero, enter - (4): (3) (4) 4, and (4) = 2^wJc(xi,Xj)- 2C(x,xi),i = 1,...,η f C(X丨·,X)) = C(X,Xy ), / 1 f ...yf7 In an embodiment of the present invention, n in equation (4) is set to expand into equation (5): 1274908 where c(x〃x)), Q system is), and finally adjacent color pixels The value of A can be forgotten; π 击二笛哲. ~w{ W2 W3 = w4 _ A _ CH c12 c13 C14 i C21 C22 C23 C24 1 C31 C32 C33 c34 i C41 C42 C43 C44 i ~ 1 1 1 l〇γ -1 'c10~ C2〇C30 C40 - 1 (6) For %, please refer to Fig. 2 and Fig. 3 for this one. A flowchart of an image color interpolation method of the example of the %, and a third figure illustrates the implementation of an image color interpolation method according to an embodiment of the present invention. Step Flow Chart Step 10: Capture the pixels of the color filter array and send 3 steps 20 to perform. Referring to FIG. 1 again, the blue pixel (Bi, j) is set to middle < ”, and the pixel is occupied, and then step 20 in FIG. 2 is performed. · Interpolating green on the blue dad pixel, the interpolation method is The flow chart is as shown in Fig. 3. The implementation method is as follows. Step U0: Calculate the average pixel as shown in equation (8): 〇α ^ + G/+UJ" 〇iJ~l + )/4 (7) Figure 2 Show in accordance with the present invention ~ : +2 Jl + ^ / + 2, 7 + +, ru + 2 = fe-U + 2 + GiJ + l + Gi + lJ + 2) / 3G · ... -1, Bu 2 + G/+1 2)/3 K, j “2, y G··” also u+G “2j+1)/3 G, (8) where the calculation of the average pixel (G,) is utilized and centered Point pixel (Bi: neighboring Gi-2, Sichuan, Gm, [2, G..., GM, j+2, Gi ji, Gm, j-2, Gi+1, j, Gi+l j+2, Gi +2 ji and the pixel value of Gi+2, .... Then, as shown in equation (9), step 12 is performed (^ calculates the color difference values of different positions to obtain ten groups of pixel color difference values, wherein Pixel 1274908 The difference is calculated by using %o, 2Ag, JB; +2; j"^^ and 'subsequently returning (6), ie;; 2:;,: called 'the weight values are in bits · "weight value" “士", ... occupies pixels (B i, j) the weight values of the four adjacent positions (up, down, left, and right) (Wi, W2, W3, D. Next step 14G: performing color model conversion to find the center (BU) adjacent The conversion values (KB) of the four different pixel positions (up, down, left, and right), and the conversion value (κΒ) are defined as Kb = g_b. The calculation of the conversion value is performed using Gi.1J, ^, Bu 1, Gi 'j + 1, G..., j, B%·, Bi, j-2, BU, Βί+2, 』 and Bi j+2. The color conversion process is as shown in equation (10). C30 =1^/ +2,7 ~5/,>HG/+2J C40 ="j+2 -'"+|G/j+2 '12 =1^-2,厂5/J-2| 啦一2 j C13 =(|Do as well-2, the factory does +2 2,y C14 APi-2J ^BiJ^+\Gi-lJ [j-2^K2j\+\Gu_2 ^24 APiJ-l and /+2, the factory is done" +2丨啦+2,7· -0,^)/256+1 _G/j 1)/256+1 Ί6+1 ～2,|)/512+1 ,()/512+1 - G, -G,y+2()/512+l ))/512+1 {/512+1 -G, , +2j \j^2 JU+2 丨/512+1 (9)

Cj! =C22 =[33 =C44 =〇,iVu) = GU,r((t2J+t)/2)^ ^ B(i,jl) ~ ~ ((^i,y-2 + ^i,j -(fe+2,y + A" V 2) (10)IW".+i) = A+i - ((to do "+2 + do "·)/ 2) Next, by equation (6) Calculate the weight value and the conversion value (Kb, value obtained by equation (10), and calculate the color estimation value via equation (11), step 150. Then proceed to step 12 1274908, step 1 60: calculate the interpolation value (Gi,j) The interpolated value is calculated using the center point pixel and the color estimate, as shown in equation (12). This interpolated value (Gij,,) is the blue-green interpolated value at the blue pixel position. = ><K:B{ijA)+W, x^B(i+lj)+W4 xiCB(ij+l) WBij+Bid (12) Repeat step 20 until all blue pixels in the CFA are processed. Go to the next step.

^Please refer to FIG. 4, set a red pixel (Ru) as the center pixel, and then proceed to step 30 in FIG. 2: interpolate the green pixel on the red pixel, this step 30 is similar to step 20, but the center pixel Not the same. The interpolation method is also shown in the flowchart of FIG. 3, and the implementation method is as follows: Step 110: Calculate the average pixel, as shown in the formula (8): The average pixel (G) of the eight is used and the center point pixel (heart)) Gw, 』", Gi_2J + 1, Gmj-2, Gm, 』, Gi'j+2, Gi'p, Gi j + 1, Gi+1, j, Gi+1, The pixel values of j+2, Gi+2, w and Gi+2J + 1. Then, as shown in the formula (13), to perform step 12: calculating pixel color difference values at different positions to obtain ten sets of pixel color difference values (Cij), wherein calculating the pixel color difference values is performed by using Ri 2 j, Ri j 2, Ri j, &, the pixel values of Bu 2 and U j and the average pixel (G,) calculated by equation (8), and then returning to (6), 'that is, step 130 : Calculate the weight value, which is the weight value W3, W4) of the four adjacent positions (up, down, left, and right) of the center point pixel (Ri, 』). Then, the step element d, j) adjacent 140: performs color model conversion to obtain a conversion value 13 1274908 (KR') at a central point image of four different pixel positions (up, down, left, and right), and converts The value (KR') is defined as kr=GR. Outside the conversion value 瞀 DT heterogeneous use

Gi-ij - Gij.! ' Gijj + 1 - Gi + 1J > Ri.2,j ' Rij-2 " Rij - Ri + 2 . ^ R ' J ^ i, j -f 2 0 and the color conversion process As shown in formula (14). c Chuan eleven 2 ~ divination. One 2, production ten ^, /-, -Gz·, /丨)/2 leak C3〇=(^+2j*-^ji^/+2j--G/,yj)/255+7 C^+ z.J>2-/?〇.|+|(^J+2f—Gz.,/|)/z^+7. /2+/-,)-'7-2丨+丨. /-2, production % - 2, |) / commit + /

. "=卜._以-~+2|+卜2,)%">2||/5/2+/ c23 ^\Ri,j-2-Ri+2j\+\GiJ-~2 ,-Gi+2,j,^l^^+i C~~_2 -/^2|+|G/,^LGi J+y|j/5/2+/ 04=^+2,)-^ +21+6+2, /-(^+/1)/5/2+/ , CU II C22 33 =C44 =0 , ( 1 3 ) ^fR(il,j) = ^i-lj ~ (( ^/-2J + ^/j)/ ^ ^R(ij-i) = ^ij-i - ((^/,;-2 + ^/j)/ ^ R(i+l,j) = ^i +JJ ^{{Ri+2j+Rij)/2) (14) K'R(U+" = GiJ+] — ((RiJ+2 + Ru)/ 2) Next, the weight value and conversion value obtained from the above (KR,) value, via equation (15), step 150: Calculate the color estimate. Then proceed to step 16〇·^10 to calculate the interpolation value (Gi,j), which uses the center point pixel and the color estimate. The calculation is as shown in equation (16). This interpolated value (Gi, "'') is the red-green interpolated value at the red pixel position. ^r,R〇J) = X^R〇-lJ)+W2 ^RO+Jj)'^ R(. } (15) (16) Until all red pixels in CFA have been processed

GUn=K,, RU Repeat this step 30, and then proceed to the next step. 14 1274908 Please refer to Fig. 1 again, set a blue pixel (Bi, 〗) as the center point pixel, then proceed to step 40 in Fig. 2: interpolate the red pixel on the blue pixel, the flow chart of the interpolation method As shown in FIG. 5, FIG. 5 is a flow chart showing a second implementation step of an image color interpolation method according to an embodiment of the present invention, and the implementation method is as follows:

Bi-2j-2 ^ BiJ + C60=( BU2j-2 - Bu + . Ton!!, '-Gq' C7.=( + C80 == \ Bi~2j+2 - Bij + Gi-2,j+2 "-GJ' 1/256 + 1 )/256 + 1 )/256 + 1 Step 210, 210: Calculate the pixel color difference at different positions, as shown in 〇7): ' Λ , ", /256 + 1 ' (17) ^ ^/+2,7-2 + ^/-2J-2 ~^/+2j-2M )/256 + 1 - h,J,2 + ^/-2J-2 ~^/+ 2j+2M )/256 + 1 - do one 2, _/+2 + ^/-2,7-2 ~^/-2j+2f, )/256 + 1 -β Ui+2J+2 + ^/+ 2,7-2 ~^/+2j+2M )/256 + 1 -A-2,y+2 + ^;+2J-2 -^/-2,7+2 ' )/256 + 1 -End - 2 J+2 + ^/+2, y+2 J+2M )/256 + 1 C55 = C66 = C77 = C88 = 0

• The calculation of the pixel color difference is calculated using Bi.m, Bi_2 j 2

Bi'j, Bi+2, j.2& Bi+2, j + 2 pixel values and blue green obtained by equation (12) • Interpolated values Gi-2'j-2, Gi-2.j. 2,,,Gu,', Gi+mi Gi+2 j + 2,,. Find the color difference of the pixel at different positions in the group (after cy, then return to equation (6) to perform step 220: calculate the weight value, which is the pixel at the center point (four adjacent positions of U) Weight values (Wi W2, w3, w4) (upper right, lower right, upper left, and lower left) Step 230: Next, perform color conversion to find the pixel at the center point (B. · four adjacent sets of different pixel positions ( Conversion of upper right, right τ, upper left and lower left) j 15 1274908

Adjacent R, R i+! , R value (KR). The conversion value (KR) of the billing center and the center point pixel...) diagonal M, j + 1, R i + + i and the formula (16) obtained 1 small i'', Gi+j + Γ and Gi + l J_ (18) The red and green interpolated value GmjY', the color conversion process is as shown in equation (18) K = G 丨, ! H a h K = Gi+JJ.j - Ri+J j_j K Hi one!, j+〇= Gm 川'一Ri j 川

KR(i+I'j+〇- G - Ri+J J+J • Next, from the above-mentioned weight value and conversion value (κό, you can find the color estimate by step 2 of the equation (i9) ( Kr,,), then perform the calculation of the pixel (Bu) and color estimation values of the center point to perform step 25: find the interpolation value (Ru ), and the interpolation value (Ri j'') is in the red pixel. The red and blue interpolation values of the position. KR(iJ) =^x^(M,-1)+^x^(,lj_I)+^ ) (20) Repeat step 40 until all blue pixels in the CFA After the processing is completed, the next step is performed. Referring to Figure 6, the present invention is not limited to this figure, and the green pixel horizontal phase may also be a red pixel. Please refer to FIG. 6 again, and set a green pixel (Gj). j) is the pixel of the heart point, and then proceeds to step 5 in FIG. 2: interpolating the red color on the green pixel, the flow chart of the interpolation method is as shown in FIG. 3, and the implementation method is as follows: Step 110: Calculating The average pixel, as shown in equation (25): 16 (25) 1274908

+^7+lj· 4 .\J+^i-2J+\ V ) ,)/3 n '/+2J~1 +i?/+ij+-Kz+2j+lf,)/3 ')+2, fc -U+2+AJ+1”+~u+2)/3

The calculation of the average pixel (R') utilizes the pixel with the center point (R"l j-2, Ri ij, Ri ij + 2, R, J 2 八 1 + 1 j, Ki+ii + 2 And the pixel values of Ri.2, j+i,, Lw, ^ and Ri + 2, j + 2 obtained in $10. 1) J + 1 Then, as shown in equation (26), step 120: Calculate the pixel color difference at different positions of the job ΐ = find the pixel color difference value (Cij) of the ten groups. The calculation of the pixel color difference is the use of Gi.2, j, Gu.2, Gi j, & . & p .,, .. : , the pixel values of J 2 and Gi+2,j and the average pixels Ri2j, Ri + 2j, R , , and Ru calculated by equation (25), Then, return to equation (6), that is, proceed to step 13: calculate 瞀1, and the weight values of J+2 are the weights of the four 'bit weights=the bottom, the left and the right at the center point pixel (Gi j). Value (Wh W2, w3, w4)., C2HGU-2-G/, j+j~-2u~j)/256fl ~, such as factory °4〇=^+2-~|+丨~+2' Μσ, -πσ.·一纠~-'』))/25"1

Cl Yong 2, factory c14 such as ~+2jy2 continent [24=";_2-~+2丨松^^ (26) cx! =C22 =C33 =〇44 =0 Then step 140 ··Color model conversion, eight ® at the center point like 17

1274908 The conversion value (Kr') of four different pixel positions (up, down, left, and right) adjacent to the prime (GU). The conversion value is calculated by using Gm/ and Gi屮u obtained in the equation (16), Gu + 1'' and Gi ji obtained from the equation (12), and the center point pixel (Gi'j). Adjacent Rkj, Ri+1′′, and the heart obtained by the formula (2〇), and RU + 1 . The color conversion process is shown in equation (27). K'r{mj) = Gi+lJ^Ri+{j (27) , ruler, (,, "ι) = G/, Chuan "- and. +1'' Next, from the weight value and the converted value (Kr,) value obtained above, the color estimation value (Kr" is calculated via the equation (28), that is, step 150: and then the center point of the equation (2^) is performed. The calculation of the pixel (Gu) and the color estimate (Kr'' is followed by step 160. • Calculating the interpolated value (Xij). This interpolated value (Xu'') is the interpolated value of the green red at the green pixel position. (28) xu':Gu-K\ (29) Repeat this step 50 until all green pixels in the CFA are processed. 0 Refer to Figure 4, set a red pixel (Ri j) as the center pixel, and then proceed. Step 60 in FIG. 2: Interpolating blue on the red pixel, the flowchart of the interpolation method is as shown in FIG. 5, and the implementation method is as follows: Step 210: Calculate pixel color difference values at different positions, such as 3〇) shown: 18 (30) 1274908 0^ 0 0 6 7 8 7 8 8 C5C6C7C8 QQ C5 Q a C7 -2,7- V+2J-2 /+2J+2 /-2,y+2 - Ri,j + -Ri,j + -Ru + G,•...Ά,, "RiJ + - GJ /256 + 1 /256 + 1 /256 + 1 /256 + 1 ^/-2,7-2 ^ /-2,7-2 -2j-2 /+2,y-2 :+2,)-2 /+2J+2 Ri+2j-2 + Gi-lJ-Ί -GiVl,卜Ί ) /256 + 1 ^+2j+2 + Gi-2,j-2"-Gi+2J+2 )/256 + 1 ^/-2,7+2 + ^/-2,y-2 - ^/-2,7+2 )/256 + 1 ^i+2,j+2 + ^i+2J-2 -^/+2,7+2 )/256 + 1 ^/-2,y+2 + Gi+2j-2'- Gi-2J+2 )/256 + 1 and,-2"+2 + ^/+2,7+2 ·~^/-2,;+2 )/256 + 1 &quot ;66 "77 "88 55 c The calculation of the pixel color difference is based on the pixel values and expressions of Ri-2j-2, Ri2j2, Ri+2, j-2 and Ri+2, j+2 ( 16) The obtained red-green interpolated values are 0 "2 small 2", 0 "2 small 2', Gi, j", Gi+2, j-2', and Gi+2 j+2'. After finding the pixel color difference values (Cij) of ten different positions, then returning to equation (6) to perform step 220: calculating the weight values, which are located at the central point pixels (four adjacent positions of U) Weight values (W], w2, w3, w4) 〇Step 230: Then perform color conversion to find the pixel at the center point (1 adjacent four different pixel positions (upper right) , bottom right, top left, and bottom left) conversion values (Kb'). The calculation of the conversion value (Kb') is based on the center point pixels (ri + i, W, d M, j + 1, B i+ij + i The blue-green interpolated value Gi-uj-Γ, Gi+1 · 1 small 1 color conversion process adjacent to + ' ' GM, j + 1 and Gi + lj + 1'' obtained by equation (12) Equation (31): 19 (31) 1274908 KR{il,j~l) = G ^ K, Hi^j~〇= Gi+1J^ K = G^J j+jK = G/+yy+;

—R —R

R -i, H , ji ~ i, j + l 'from the above-mentioned weight value and conversion formula (32) step 240 • • find the color estimate (Kb,,), lack, and two by juice τ center point pixel (U and color estimation value, for example, (33) out of interpolation value, this interpolation value (Bij,,) is in the ^ □ / 〇 25 〇: color red interpolation value. ', " For the blue pixel position, the blue K\id, =^x^, _w_1)^2 ΧΙ0Β(Μ.^ xKB(iAJ+l)+W4 χΚβ(^ (32) v 鸣,' (33) repeat In this step 60, until the processing of all the red pixels in the CFA is completed, the next step is performed. Referring to FIG. 7, the present invention is not limited to this figure, and the green pixels may be horizontally adjacent to the blue pixel, please refer to the seventh. In the figure, the green pixel is set as the center pixel, and then the step 7 in FIG. 2 is performed: the blue pixel is interpolated, and the flow chart of the interpolation method is as shown in FIG. 3, and the implementation method is as follows: Step 110 : Calculate the average pixel, as shown in equation (34): fly, /, now,) /4 5/-2, y. J-1"+1:1 j + Bi -2J+1 丨,)/3 (34 ) <5,+2j'=(A+2,/+A+U+5,+2j+l")/3

BiJ+2 f==fe-i5y+2 + Bij+\ ^BMj+2 )/ 3 ; 1274908

The calculation of the average pixel (B') is advantageous for the adjacent Bi-丨i.2, R. Ώ not used. Point pixel (Gu) 'Bu 1'·1, Βμ,·ϊ + 2, Bi + 1 j-2, Bi + 1 j, B (33), Bi2M, ' ' , j+2 and The pixel value of Bi+2j + 2. i, j + i difference = = 3:) shown in step 12. : Calculate the pixel color of different positions by using G—i”, .·, :), and the pixel value of the medium-wide pixel color difference 4, the correction of the if M Gi, and the pixel value of Gi + 2,j (34) The calculated average pixel Bi 2 j,

B i + 2,j

B i J-2

B, j + 2

, , J - L /," and then return to equation (6), that is, step 130 is performed. · Calculate the weight value, which is the four adjacent positions of the center point pixel (Gy) (upper, -ΤΓ, Weight value of (1, W2, W3, W4) 1+1^/+2, /-5^/1)/256 +1 C2〇= Ip; y_2 -g c3 . =φ,.+2 r G/ C4〇=fG,.+2-GΗ,"-. . ·-: ...'G i C,4 =f〇^2 .-G,.y+2 C ' c

C C13 = φ, 2, wide (ii + 2 ) · i, j + ‘ ii, y-2 -^/ + 2,

iG C 34 = ^1+1^/,^2^5,/1)/256 +1 Ten,-2,/-^:2.1)/256 +1 BubΌί+2 /丨)/256 +1 + \B»-2,J'-Bi j+2^256 +1 |5,·.;ϋ + 2"·|)/256 +1 \B^j-2-Bitj+2^/256 +1 + \Bi^i,j-Bi J+2^/256 +1 (35) C11 =c22 =c33 =c44 =〇Next step 140: Perform color model conversion to find the pixel adjacent to the center point (Gu) The conversion values (κΒ) of the four different pixel positions (up, down, left, and right). The conversion value is calculated by using Gi l j'' and Gi + 1, j'' obtained in the equation (12), and GiJ + 1'' and Gi obtained from the equation (16), BM, j, Bi+1, j adjacent to the center point pixel (Gu), and Bij-1, and Bi, j + i obtained from the equation (34). The color conversion process is as shown in equation (36). 21 (36) 1274908

= Gi-lJ Ίζ I B. -i,y Next, from the weight value and the converted value (Kb') value obtained above, by (37) step 150: calculating the color estimate (Kb',), Then, calculations such as the center point pixel group and the color estimation value (Kb" are performed, and then the step 'calculate the interpolation value (Xi j)). This interpolated value (~Γ) is interpolated in green blue at the green level. . μ Κ, 财 哨(四)年 '~纲(37) xid =GK"KBiJ (38) Repeat this step 7〇 until all green pixels in the CFA have been processed = the preferred embodiment of the present invention described above, the present invention The feature is to interpolate and restore the color image of the color filter and the light P train by estimating the minimum flat error. In order to estimate the image value lost in the color filter array image, the weight of the near color pixel is: The value is determined by the matrix calculus. A color model (KR ' κ Β ) is also used in the process to achieve a better image effect. Although the invention has been disclosed above in a preferred embodiment, it is not intended to limit the invention. 'Anyone skilled in the art can make various changes and refinements without departing from the scope of the invention. Therefore, the scope of protection of the present invention is defined by the scope of the appended patent application. The above and other objects, features, and advantages of the present invention will become more apparent from the description of the appended claims. Diagram A schematic diagram of a color filter array is shown in FIG. 2 is a flow chart showing an image color interpolation method according to an embodiment of the present invention. FIG. 2 is a diagram showing an image color according to an embodiment of the present invention. FIG. 4 is a second plan view showing an image color interpolation method according to an embodiment of the present invention. FIG. 5 is a second plan view showing an image color interpolation method according to an embodiment of the present invention. FIG. 6 is a third plan view showing a CFA according to an embodiment of the present invention. FIG. 7 is a fourth plan view showing a CFA according to an embodiment of the present invention. Component Symbol Description Step 1 〇: Capture the pixels of the color filter array Step 20 · Interpolate the green on the blue pixel Step 30: Interpolate the green on the red pixel Step 40: Interpolate the red on the blue pixel Step 50 ··Insert red on green pixels 23 1274908 Step 60: Interpolate blue on red pixels Step 70 · Interpolate blue on green pixels Step 11 0: Calculate average pixels Steps 120, 210: Calculating pixel color difference values at different positions Steps 130, 220: Calculating weight values Steps 140, 230: Performing color model conversion Steps 150, 240: Calculating color estimation values Steps 160, 250: Calculating interpolation values 5: Color filter array

twenty four

Claims (1)

1274908 X. Patent Application Range 1. An image color interpolation method, comprising: providing a checkerboard array having a plurality of odd-numbered columns and a plurality of even-numbered columns arranged in a staggered manner, wherein the checkerboard array is provided with a plurality of first pixels, plural numbers a second pixel and a plurality of third pixels, wherein one of the third pixels is staggered with the first pixels on the odd columns, and the other of the third pixels and the second pixels Arranged on the even columns, wherein each of the third pixels is horizontally and vertically adjacent to one of the first pixels or one of the second pixels, respectively The first pixels have a first primary color value, a first second color interpolation value, and a first third color interpolation value, each of the second pixels having a second primary color value and a second first color a color interpolated value and a second first color interpolated value, each of the third pixels having a third primary color value, a third first color interpolated value, and a third second color interpolated value; performing a first interpolation Steps to target these a first target pixel of a pixel, using twelve third pixels of the third pixels that are most adjacent to the first target pixel, and horizontal or vertical phase of the first pixel with the first target pixel Having four first pixels adjacent to calculate a first third color interpolation of the first target pixel; performing a second interpolation step to utilize the second target pixel for the second pixels Calculating the twelve third pixels of the third pixels that are most adjacent to the second target pixel, and the four second pixels of the second pixels that are horizontally or vertically adjacent to the second target pixel. One of the second target pixels is interpolated by a second third color; 25 v/4908 performs a third interpolation step to align some of the second pixels with *(four) pairs of (four)-target pixels, using two pixels, The first pixel value %# adjacent to the pixel diagonally adjacent to the first pixel, and the second third interpolation value and the first value, the Γ-t second pixels The second third interpolation is diagonally adjacent to the <the first to third interpolated values and the first target The first of the pixels = the first pixel of the pixel position of the — - the interpolated value, and the philosophical. The pixel is the most adjacent four of the second one: the pixel in the prime and the first target The first-second color interpolation I is used to calculate the first target:: two, the interpolation step 'for the third pixel-the third one' is the most common among the second pixels Six pixels adjacent to the pixel of the third target, and the pixels of the first one are most adjacent to the pixel of the first pixel of the first singular diom interpolated value The first first color interpolated value in the first pixel, the two first color interpolated values of the first one color of the first pixel, and the two first pixels adjacent to the second horizontal pixel The second third color interpolated value of the second second pixel of the pixel, the second third color interpolation of the second pixel, and the second second color of the second pixel vertically adjacent to the third target pixel And the fourth pixel of the second pixel that is horizontally and vertically adjacent to the third target pixel to calculate the third target pixel - a second second color interpolation value; a fifth interpolation step, for the second target pixel, using the fourth pixel adjacent to the diagonal of the second target pixel in the first pixel The first third interpolation 26 1274908 value of the four first pixels adjacent to the second target pixel diagonal line of the first third interpolation value of one pixel, and the second second of the second pixels The second interpolated value of the five second pixels in the three interpolated values adjacent to the second target pixel and the second second pixel in the second target pixel position, and the second target and the second target Calculating a second first color interpolated value of one of the second target pixels; and a fourth second pixel adjacent to the diagonal of the pixel; ^ 仃一弟六 interpolation step, for the pixel of the third target, using the six first images of the pixels adjacent to the third target of the first pixel: the second of the two pixels The first color interpolated value is the most adjacent to the sixth, and the second first color interpolated value in the second pixel is adjacent to the horizontal. , the _ two third color interpolated values and the third target pixel - like the second third color interpolated value of a first pixel, the two adjacent _th pixels - ^ The pixels in the vertical pixel, the μ first interpolated value, and the third element:,: the four third images that are perpendicular to the third target pixel are different from the third The third-numbered pixel-third third-color interpolated value. Method, complex medium = green _ patent range of the image color interpolation described in item 1 of the first rod of the first rod = the first target pixel is ^, and Γ. IsJ'2 Gi-l, j ' Gi.lj + 2 > G. ^ 1 + 1,j, Gi + 1,j+2, G 1,J + 1 丨+ 1 small 2 The first target is like Xiao Xiao and Gw, 』+〗, The first pixel is horizontally or vertically adjacent to the pixel, and the second pixel is 艮_2, and the wide interpolation value is G., /+2. a third color I, J, the first interpolation step includes at least 27 27 274 908 to perform an averaging operation to obtain a plurality of average pixels (g,). The averaging operation is based on a formula: where G'-d, -丨(G/-i,y+2 +川+ G/+u+2)/3 (GMj-2 +^, yi +G/+1j_2)/3 丨-u iJ+\ / Perform a color difference calculation To obtain a plurality of pixel color difference values (Ci), wherein the color difference calculation is based on a formula: C1. Ten / 彳 factory to the end "" cHv2 - ~ 丨 + | heart 2 C30 = ([5 / + 2, factory '" + 2 " C4 〇 = ky + 2-^ | H-| G / J + 2 CiH5/-2 "~Ί|(^_2,; Cl4 Ηαή,. j+2|+|g,'G+odjI+h -2,y U~2 C24=(b,.^^5.. +2|+|G.._2 C34=|B,+ Λ/256+1 -^()/256+1 /256+1 /256+1 JUj /512+1 -G,.j+2,| ) /512+1 -G/+2j|)/512+1 J/j+2 ?/J+2 丨)/512+1 ,丨)/512+1 ^2j-^.j+2|+| g/+2j 1Ch=C22=:C33=C44=0 performs an inverse matrix operation to find four weight values W1, W2, ^ and W4, and the inverse matrix operation is based on a formula: ~Wf W2 W3 = w4 ^ λ C11 c12 c13 c14 ί -1 cl(i C21 C22 c23 c24 1 c20 C31 c32 c33 c34 1 c30 C41 c42 c43 c44 1 c40 1 1 1 1 0_ 1 Perform a color model conversion to calculate the complex conversion The value 28 1274908 (Κ B), where the color, the model conversion is calculated according to a formula: Ά('-υ) = - (Bu 1 + square, ")/2) BUJ-D GU-\ - + Bi , j) l2); BU 5 (/, y + l) color estimation value is performed (where the calculation of the estimated value is based on a formula: performing an interpolation value calculation to find the pixel of the first target The first third color interpolated value is calculated according to the formula: 3. The image color interpolation method according to claim 1, wherein the second target pixel in the second interpolation step is Ri, the third pixels most adjacent to the second target pixel are G Gw'w, km, Gi_l j+2, Gi ji, Gi j + i, ^+ two, Gi + i, j, Gi +丨, j + 2, Gi + 2, 】.! and Gi+2, j + 1, the second pixels ~Ri with the second level or perpendicular, and the second of the second target pixels The third color interpolating value Gy, and the second interpolation step comprises at least: performing an averaging operation to obtain a plurality of average pixels (G,), wherein the averaging operation is according to a formula: 29 1274908 ?/7u+g /〇+gu-i+gu,)/4 =\pi-2J-l + Gi-lJ + G/-2J+i)/3 GW= ={Gi+lJ + G/+2,y_i + Gi+ 2J+l )/3 ϋ ={g^ij+2 + GiJ+l + Gi+lJ+2)/ 3 .GU-2Z =ipi-U-2 + GiJ-\ + GMj-2 V 3 Performing a color difference operation to obtain a plurality of pixel color difference values (Cij), wherein the color difference value operation is based on a formula: 2-~|, for example, 7-2'-~'|)/condition +/C3〇 =\Ri+2J-RiM〇MC,f~+2 -' Don't ~.+2'- , to +Bu 2, Factory Zhiqiao|+b_2,/-G/+2,/|)/5/2 +/ Factory and υ+2|+|θ-2,/-~·+/|)/still +i C23=h)-2-A+2』+| ~-~巧'|)/ still +/ Q 一~-2- and O.+d+l%—2'-G^+2'|)/5/2+i C3f| Know 2, Guanghe 〇.+2|如+和/-6^ +/|)/5/2+/ Cu=C22=C3s=C44=〇; Perform an inverse matrix operation to find four weight values W!, W2, W3, and W4, and %+%+%+% =i, where the inverse matrix operation is based on a formula: w{ 'Cn Cl2 Cl3 c14 ί -1 ~cl〇' W2 C21 C22 C23 . 24 1 c20 w3 two C31 C32 ^33 c34 1 C30 w4 C41 c42 c43 c44 1 c40 1 1 1 1 0 1 Perform a color model conversion to calculate a complex conversion value (K'R), where the color model is converted The calculation is based on a formula: 30 1274908, Guang Er called Bu + 'j/2) ^=g^A{k2j+ru)/2); h ", continent = wide (few + i〇/2) color Estimate the value of the operation to obtain a pontoon value (heart)' where the estimate is far from the openness < In addition, the second interpolation is to interpolate the first color of the second parent to the pixel of the second division of the younger brother. The pottery + Zenghe > eight impulses are based on a formula: G heart Ά. 4. If the patent application scope method is the first method, in which the image color interpolation described in the first dry stop item is H-the first target pixel in the interpolation step is B.. The second pixels adjacent to the diagonal, '1, j heng t-angle are i+i, ji, R μ i + 1 and r uj-i R and the second interpolation at the first target is G , a prime of the first Gi, j, and the first target like the symplectic pair 0 + 2 ^ "2, j + 2 and the - t % 4 diagonal adjacent to the fourth second brother The interpolation value of GM, jY, Gi + i , , , g., and the pixels of the first target are Ν〇ί+1^+ι ? Bi...2, Β·角/最相The adjacent -th pixels are first first-prepared 1+2 small 2 and Bi+2, j+2, and the first target image has a value of 1 (four) of ~' and the third interpolation step is at least = The I difference is assigned to obtain a plurality of images, wherein the color difference operation is based on a formula: plain color difference (Cij), 31 1274908 C50=( Bi-2j-2 - BiJ + "wide" G /—2上2 七/J Qo ~ ( ^/+2 J-2 ~ + 『广" Gi+2J-2 Q〇= ( BMJ+2 - Bid + wide π π G/+2j+2 - Q 〇 =( Bi a 2j+2 - Bu + . Bu 2, Bu 2 -hj 11 11 11 11 + + + + 6 6 6 6 5 5 5 5 "56 ^57 "58 "67 ^68 ^78 \ ^i-2J-2 ^'+2,7-2 + '_2,,-'female 2" \^i-2J-2 ^i+2,j+2 + one 2,j-2 -Gi+2 , j+2 one 2 - B Bu 2j+2 + [Bi+2j-2 - Bi+2j+2 + Gi+2J.2^Gl+2J^ [Bi+2J-2 - Bj-2J+2 + - Gu,;+2" {Bi+2J+2 - A-2J+2 + A2,y+2"-'+2" /256 + 1 /256 + 1 /256 + 1 /256 + 1 /256 + 1 /256 + 1 C55 = C66 = C77 = C88 = 0 Perform an inverse matrix operation to find four weight values Wi, w2, w3, and w4, and %+%+%+% = /, where the inverse matrix operation According to a formula: Wf_cn Cl2 c13 Cl4 1 -1 -c10- W2 C21 C22 C23 C24 1 C20 W3 = C31 C32 ^33 C34 1 c30 w4 C41 C42 c43 c44 1 c40 _ λ _ 1 1 1 1 0_ 1 A color model is converted to 9 (K'R), where the calculation of the color model is based on a formula: K t R{il,jl)= // G " ~ Ri-ul K t R(i + ] ,jj)= gw/'- -/? K t R(il,j+l)= -RK 9 R 0+^J+O ~ Gi+u+i - _ R Perform a color estimation calculation to Get a color estimate ( <), wherein the calculation of the estimated value is based on a formula: 32 1274908 ') =_1) + ❿乂 (four)) + heart _ ^ into the calculation of an interpolation value to find the first pixel of the first target The second color interpolated value is calculated according to a method of a female female H/-K, and Ri ... A. 5. The method of image color interpolation according to the middle of the patent, wherein the fourth interpolation The third target pixel in the step is called, and the second pixel corresponding to the pixel of the S triple mark is 1, J 2 Rl+1, j-2, R "i, j+2, and Ri+ 1, j+2 'the first-second color interpolated value among the six first pixels adjacent to the first pixel of the third target is: η·2,... The second color of the first-color is: the interpolation ', ' and Gi, j+1' are vertically adjacent to the pixel of the third target -, and the second third color interpolation of the two pixels of the Gudi And: 'fourth horizontally and vertically adjacent to the third target pixel! t: 2'j, Gi, j-2, Gi+2, j ' and Gi, 』 + 2, the third target pixel #弟一弟—The color interpolation value is Xij”, and the fourth interpolation step is at least two people: An averaging operation to obtain a plurality of averages i3. The averaging operation is based on a formula: ,, ()' where 7?.. i, J + rmj+Rij^+Riy ψ") / /? · ^ .= 1 - Shape - 2,1 "+ and -i"· + and /—2, State 1 ή/3 ^/+2,7 = ")/3 Bu 2'= )/3 (^/-l,y +2 +^/,y+l ?'+^/+l,y+2 . )/3 33 1274908 Perform a one-color difference operation to obtain a plurality of pixel color difference values (Cij), where the color difference calculation system According to a formula: . 10 10, 1 2, the factory is 妒56+1 C40=(Gu+2-Gj 啦"+2,-i^;|/256+l -12 =|^/-2j_^/, y-2| +|^-2j,~^J-21)/^56+1. 13+_2, factory 0/+2, 』啦_2, production rule..)/256+1 C14 =ί^_2,) -(3^.+21+17^2/-^+2 j) /256+l ^23=i^j-2 ~^/+2,y |+|^·,/-2 ~^r+2j· j)/^56+l · C24=j~-2-, 2|+|A_2,-~.+21)/256+1 'C34+,+2,factor~2|啦+2,/-1+2|)/256+1 qj =C22 =C33 =C44=0 Perform an inverse matrix operation to find four weights (Weights) W!, W2, W3, and W4, and %+%+%+%=/, where the inverse matrix operation is based on a formula: wf ~cll Cl2 C13 c14 f —1 -cw W2 C21. 22 C23 C24 1 C20 w3 = C31 ^32 ^33 . 34 1 C30 w4 C41 C42. 43 c44 1 c40 A _ 1 1 1 1 0_ 1 Only one color model conversion > to calculate (K'R), where the calculation of the color model conversion is based on a formula: KG丨心, -lu K' 丨一a Gu-r KR{i + Kj)= G, _+w K{ ^(/,7 + 1)= 〜+r Perform a color estimation operation to obtain a color estimate (〇, where the estimate The calculation of the value is based on a formula: 34 1274908. The calculation of the interpolated value is performed to obtain the third 1st and second color interpolated values, and the calculation is based on the formula: the 1 factory method of the prime, the middle 嗲筮X 私私止丨< The second target pixel in the r-interpolation step of the sakura interpolation is the R δ hai ji two-labeled back disk & Μ二^, which is Rlsj ' ” two::1 ' The second ^ G i+2, j·2, G i+2, j+2, ', G i 2 of the second pixels of the pixel position of the pixel: 2, the four pixels of the second target diagonally adjacent to the first pixel-interpolated value G i , .,,,, r,, 豕常·十十·*·1, G ...'·Μ, GMj + 1 ',, brother-labeled pixel diagonals are the most adjacent to the first Like Ri+2, j·2 and Ri+2, j+2 'the second target pixel two j value is Bij''' and the fifth interpolation step is at least ^ operation 'to obtain a plurality of pixel color difference values ( Cij) L transports according to a formula: i+l, jl, Β μ," and the second interpolated value Gi,j, and the first third i+l,j + l , and the AR "2, j-2, R The second first dad contains: έ value -2,;;? 2,72j a difference ^ line color 50===60==0==0== into the ic丨C1C7C8 Φ—its -2-R 1 1 . 丨+ g-2,^2m-g. + °W2" 乂·” + hG:, / 35 1274908 C56=( ^/-2,7-2 - ^/+2,7 -2 + Gi-2j-2 -Gi+2J_2 C57 =( Ri-2,j-2 _ Ri+2,j+2 + G/_2,/_2''-G/+2j+2” C58 =( ^/-2 J-2 ~ ^/-2,y+2 + < C67 =( ^/+2,7-2 ~ ^i+2J+2 + ^/+2J-2 ~^i+2, j+2 Qs =( ^i+2J-2 ~ ^/-2, y+2 + C78=( ^/+2,;+2 — ^/-2,7+2 + Gi+2J+l -Gi -2,j+2 C55 = C66 = C77 = C88 = 0 lx IX 11 11 11 1Χ + + + + + + 6 6 6 6 6 6 5 5 5 5 5 5 Perform an inverse matrix operation to find the weights The values Wi, w2, w3, and W4, and wherein the inverse matrix operation is based on a formula: Ί7 _cw C20 C30 c40 l Wj Cn cn cl3 cl4 l W2 C21 C22 . 23 . 24 l W3 = C31 C32 c33 C34 l w4 c4l c42 c43 c44 l λ llll 0 Perform a color model conversion to calculate a plurality of conversion values (Κ'β), where the calculation of the color model conversion is based on a formula: r 丨" K = G » &quot ; K = Gi+l J_{ — Bi+l j_x , KB(ii,j+\) = G(_1υ+ι -» " K 5(/+1,y + l) = G i+lJ + l — Bi+lj + l for a color estimate Computing to obtain a color estimate (wherein the calculation of the estimate is based on a formula: performing an interpolation calculation to determine the second first color interpolated value of the second target pixel, the calculation is based on A formula: a ~. 36 ^ 274908 7 · If you apply for the scope of the patent, the law of the first paragraph, in which the first 丄 芈 芈 _ ^, the description of the shirt like the color of the interpolation you, 内, insert step 申申誃The pixel adjacent to the first:r_7^ is Gj j , and the six first B-W^B, 1J+2^Bi+i. pixels of the pixel of the most z-one private pixel are in the second pixel of the pixel. The first color interpolated value of Bi^,, Β, , ,, and the third target is the same as the first I, j+1, Bi + 2, _M, and Bi + 2 , j + 1'', the pixel-level adjacent two-color interpolated values Gi_, ', and Γ, the first 苐Μ·丨 and G... of the material, perpendicular to the third target pixel One of the neighboring brothers like Xin Zhiqin Feng Feng knows that the brother of the brother-in-law is a two-color interpolated value of Gi•丨, the factory and 1+1'j′ are horizontally and vertically adjacent to the third target pixel. First The pixels are 2, j, Gi, j-2, Gi+2, j and Gu+2, the first color interpolation of the third target pixel is Xi, j, and the fourth interpolation step includes at least : performing an averaging operation to obtain a plurality of average pixels (B, wherein the averaging operation performs a color difference operation according to a formula: )/3 to obtain a plurality of pixel color difference values (CJ, wherein the color difference value The calculation system is based on a formula: 37 1274908 c.Ag^jg^ ~丨)/25^1 Q〇=|G/+2j_G/j|+|^/ +2J , for 1)/256+1 M~2| +一^ C^G^jG^ Ch+^-g^I+I^ C23 =^"_2 -0/+2"|+|Do"_2'-5/+2"j/256+1 C2 , 4pij-2-GiJ+^^^^^^ Q j =C22 =C33 =C44 =0 Perform an inverse matrix operation to find four weights (Weights) Wi, W2, W3, and W4, and) = 7 , wherein the inverse matrix operation is based on a formula: wf _. 11 c12 c13 c14 1 a 1 _Cw w2 C21 C22 c23 . 24 1 c20 w3 = C31 C32 C33 c34 1 C30 w4 C41 c42 c43 c44 1 C40 again 1 1 1 1 0 1 Perform a color model conversion to calculate a plurality of conversion values (K'b), wherein the calculation of the color model conversion is based on a formula: JT» - = G λ n- - β . ' ϊ-1,7 Y 1一^一一〆- -B . /1 Ή · fC1 - -B, ' ·, ϊ+1,7 Y 1 Λ B(i,j+l) one-Β χ1π ϊ,7 + 1 for a color Estimating the value to obtain a color estimate (〇, wherein the operation of the estimate is based on a formula: 38 1274908 - calculation of the interpolated value to find the third first color of the third target pixel Interpolation, the calculation is based on a formula: XUJ. The color interpolation of the image is the color of the color of the younger brother. The color is the red color. The first color is the third color of the blue color. 9. For example, if the patent is applied to the Fan Garden Law, the first color is the third color of the image color interpolation described in the item. The color is blue, and the color is blue.