TW201230792A - Method of detecting a bad pixel of a sensed image - Google Patents

Abstract

A method of detecting a bad pixel of a sensed image is disclosed. The difference of a detecting pixel value and neighboring disposed pixel values is determined. The continuity of the detecting pixel value and neighboring same-color pixel values along a specific direction is then determined. The smoothness of the detecting pixel value along the specific direction is determined.

Description

201230792 VI. Description of the Invention: [Technical Field] [0001] The present invention relates to an image processing, and more particularly to a method for detecting a bad pixel of an image. [0002] 先前 [Prior Art] Each pixel of a color image can be represented by a vector containing three color components, for example, using the light intensities of the red, green, and blue bands in the visible spectrum. In order to reduce size and cost, in a typical image sensor, only one optical filtering unit is used for each pixel to capture a single chromatic value. Then, another color value is obtained by interpolation according to pixel values adjacent to the same color. [0003] The figure shows an image sensor array 10 of image sensors covered with a color filter array 12 . Each color filter (CF) of the color filter array 12 allows only one color of light to be sensed by the corresponding light sensing element. Since the special mosaic pattern of the color filter array to 12 is not a complete image viewed by the human eye, it is necessary to reconstruct the image that can be viewed by the human eye by color interpolation. Therefore, color interpolation is also commonly referred to as CFA interpolation, color reconstruction, or demosaicking. The color filter array 12 illustrated in the first figure is a Bayer filter which is commonly used in digital cameras or cameras. The Bayer filter pattern is arranged by the red (R), green (G), and blue (B) filters as shown in the figure, which contains 5 〇% green (◦), 2 5 % red (R) , 25% blue (B); wave filter. The other color at each pixel location can be obtained from the surrounding pixels. Even Bayer pattern interpolation is commonly used in the field of image 100100060 Form No. A0101 Page 3 / Total 26 Page 1002000107-0 201230792, and it has some shortcomings, ~ aii asing Now selling, A a color shirt phenomenon caused high frequency details to disappear the new frequency and seriously damage the image.吾屋生 [0005] In addition, the film (10) "will have a bad material bad pixel." The silk is made by you, _>_ 农" £,

It is impossible to effectively distinguish the edge image of the bad point, so the 坏I 坏 误 误 误 误 误 误 误 误 误 误 误 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘 边缘In this way, image details can be preserved when compensating for dead pixels. [Invention] [0007] In view of the above, one of the objects of the embodiments of the present invention is to effectively distinguish between a dead point and an edge shadow to correctly perform a bad point correction or compensation like the == method. [0008] In the embodiment of the present invention, the difference between the value of the pixel to be detected and the value of the complex adjacent bit is determined. Then, the connectivity of the pixel value to be detected and the pixel value of the plurality of adjacent color pixels in a specific direction are detected. - After a specific direction, the smoothness of the pixel value in the specific direction according to the specific direction of the shape. When the pixel value to be detected is determined to be a dead point, the pixel value of the same color pixel may be further used according to the complex point of the dead point. Compensation for Bad Points 实施 [Embodiment] The second figure shows a flowchart of a method for detecting a bad pixel ("4 Pixel") of an image according to an embodiment of the present invention. This embodiment can be applied to various image sensors, such as complementary metal oxide semiconductor (CM〇s) image sense 100100060 Form No. A0101 Page 4 / Total 26 Pages 1〇〇2〇〇〇ι〇γ^ 201230792 Device. In this embodiment, the sensed image values for each pixel are obtained via a Bayer color chopping array. If a bad pixel (abbreviated as "bad point") is detected, the dead pixel can be further corrected or compensated. After the corrected dead pixel value is output, it can be stored, displayed or further processed. [0010] In step 21, the difference between the pixel value to be detected and the pixel value of the adjacent position is examined, wherein the neighboring pixel material and the pixel to be (four) are different colors. In this embodiment, if the pixel to be detected is centered, ο will form a 3χ3 array with the adjacent pixel, as shown in the third figure (8) the pixel to be detected touches the green (g), red (1) The neighboring pixel values are N0-N7. [0011] The fourth figure shows the detailed flow circle of step 21. First, in step Sichuan, obtain the difference between the (4) and (4) and the "material position and the N6 average value" as the vertical neighboring metric value. If the difference is less than 〇, then the vertical neighboring metric value is ^Yang Step (1) can be expressed as

BMl=Nc-(Nl+N6)/2 if BMl<〇 then BM1=0.

[0012] At / step 212, obtain the difference between the pixel value to be detected Nc and the water value N3, the level of the data, and the 豕 w w = as the horizontal neighboring metric ΒΜ 2 ❶ where the value is less than G, then the horizontal proximity The metric value Μ2 is G. This step Z 1 Z 1 is expressed as follows: BM2=Nc-(N3+N4)/2 if )n if BM2<〇 then BM2 = 0.

[0013] Step 213, obtaining the difference between the pixel value to be detected and the average value of the four corner-to-close position pixel values M〇, N2, \IG XT*7*r N5, N7 as the corner proximity 100100060 Form No. A0101 Page 5 of 26 1002000107-0 201230792 BM3. Wherein, if the difference is less than 〇, the corner neighboring metric value BM3 is 0. This step 213 can be expressed as follows: BM3 = Nc - (N0 + N2 + N5 + N7) / 4 if BM3 < 0 then BM3 = 〇 [0014] [0016] [0017] Next, in step 214, according to the vertical The neighboring metric value BM1, the horizontal neighboring metric value BM2, and the corner neighboring metric value BM3 are obtained to obtain a comprehensive adjacent sound volume value BM4. Wherein, if the value is greater than the pixel value to be detected Nc, the syndrome neighboring metric value BM4 is equal to the pixel value to be detected Nc. This step 214 can be as follows: 'BM4=(BMl+BM2)/2+BM3 if BM4> Nc then BM4, Nc Finally, in step 215, it is determined whether the absolute difference of the integrated neighboring metric value βΜ4 pixel value Nc is a small neighboring metric. The critical value β, Τη = is determined as S 'the material detection pixel material can be _. The right part of this step 2 is as follows: Table if ABS(10)4-Nc)<BjrH_& may be a bad point. The above steps 211-215 are described with the pixel value to be detected as blue, ", these steps are also applicable. In the case of red (1) when it is to be checked, such as the red 2 in the third ,, the pixel value of the adjacent position in the green (G) and blue (8) check is just 〇-N7. No _ place is only __ Xie replacement Bu_, 1B_TH is replaced by R_TH. When the pixel value Nc is green (G), such as the third C map or the third. The green pixel to be detected Nc and the blue water = neighbor 'in the horizontal graph The green pixel to be detected in the third yang is red and the neighbor is replaced by the (4)-deletion to 100100060. The form number A0IO1 page 6/26 pages 1002000107-0 201230792 GM GM4, Β_ΤΗ is replaced by GBjh, and step 214 can be expressed as follows: GM4=GM1+GM2+GM3 if GM4>Nc then GM4=Nc.

[0018] For the case shown in the third D diagram, BM Bu M4 is replaced with GM GM4, β_ΤΗ is replaced by GR Τ ίί, and step 214 can be expressed as follows: GM4=GMHGM2 + GM3 if GM4>Nc then GM4 = Nc .

[0019] Ο Next, returning to the flow shown in the second figure, in step 22, the continuity of the pixel value to be detected and the adjacent color pixel value in a specific direction (c〇n — tinuity) is examined. This step 22 can reduce or avoid the misjudgment caused by the pre-step 21. For example, if there is a 7-streaked (10) bright line through the pixel to be detected, the pre-detected pixel may be misjudged. The value Ne is a bad point. n By performing this step, if the green scale of the pixel to be detected is light, the material can be a bad point, and (4) is not a bad point. In the present embodiment, 'if the pixel to be detected is centered, then the adjacent pixel of the same color will form a 3χ3 array, as shown in FIG. 5A. The blue (B) pixel value to be detected is 1^. D0-N7.

[0020] In the present embodiment, the material direction is the Yongping direction (H), the vertical center, the upper right direction (NE), and the upper left direction (NW). In order to examine the characteristics of the Δ疋 direction, this embodiment uses the enhanced weight pair to view the desired side. Strong its pixel value. The sixth A@ respectively shows the Liningping direction enhancement weight (-, the vertical direction enhancement weight (B_V), the upper right direction enhancement weight -E) and the upper left direction enhancement weight (B-胛). [0021] The seventh step of the detailed step of displaying step 22 is performed by B-Η, and the weighting of each line image is performed as follows: . First, in step 221, 'the absolute sum value is taken. This step 100100060 Form No. 1010101 Page 7 / Total 26 Page 1002000107-0 201230792 BH_1=ABS[(-1)*d〇+2*D3+(-1)*D5]; BH—2=ABS[(-l) *Dl+2*Nc+(-l)*D6]; BH_3=ABS[(-l)*D2+2*D4+(-l)*D7].

[0025] Next, in step 222, the absolute sum value of each row is weighted using B_V, and the absolute sum value is taken as a continuous metric value in the horizontal direction ^-Η. As for the continuous measure Β-ν in the vertical direction, the continuous measure Β_ΝΕ in the upper right direction, and the continuous measure Β_· in the upper left direction, it can be obtained according to a similar principle. This step 222 can be expressed as follows: B_H=ABS[(-1)*BH_l+2*BH_2+(-l)*BH-3] Next, in step 223, a continuous metric value in the horizontal direction β-Η, vertical direction is obtained. The continuous metric Β—ν, the continuous metric Β—ΝΕ in the upper right direction and the continuous metric BJf in the upper left direction: f is the maximum or minimum value of the application as the continuous metric B ( X, γ ). Finally, in step 224, it is determined whether the integrated continuous metric Β(χ, γ) is greater than the continuous metric threshold B_Line_TH. If the determination is YES, the pixel to be detected has a discontinuity. Steps 223 and 224 can be expressed as follows: B(X, Y)_max=max(B_H, B_V, B_NE, B_NW); B(X, Y)_min=min(B_H, B_V, B_NE, B_NW); B(X, Y)=select_max_min?B(X,Y)_max:B(X,Y)_min; If B(X'Y)>B_Line_TH then Nc has discontinuity; wherein the variable select_max_min is used to select B ( X, Y) _maxi B (X'O_min. If B(x, Y)_min is selected, more image details are generally retained, but it is easier to misjudge. The above step 22 224 is blue with the pixel value to be detected Nc. (B) Time 100100060 Form No. A0101 Page 8 / Total 26 Page 1002000107-0 201230792 [0026] [0028] 1000 [0028] 100100060 2 For the description ' _ ' These steps are equally applicable to the pixel value to be detected k '' ', the case of red (1)' as shown in the fifth B diagram, the red (1) pixel value to be detected Nc and b in the adjacent homochromatic image symplectic variables are replaced only by the fact that although the fifth A picture is also applicable to the pixel to be detected The value Nc is green=, however, since the number of green pixel points is blue or red like ..., the green (8) pixel value Nc and the adjacent color pixel value shown in the fifth C-picture can be used. a map The horizontal direction enhancement weight (U), the vertical direction enhancement «(Β_ν), the upper right direction enhancement weight (β-NE), and the upper left direction enhancement weight (B-NW) are also applicable to the case where the pixel to be detected is green (8). You can also use the horizontal direction enhancement weight (G~h), the vertical direction enhancement weight (G_v), the upper right direction reluctance weight (~G-NE), and the upper left direction reluctance weight (G_NW) as shown in the figure. 'Back to the flow shown in the second figure, in the body _3, according to the specific direction passed to the previous step 22, the pixel value to be detected is detected in the specific direction π H by performing this step 23 'if The pixel value to be detected Nc is non-smooth, the shot is determined to be a dead point, and (4) is a bad point. The eighth figure shows a detailed flowchart of step 23. First, the edge value of each specific direction is calculated in step 231. As described above, the specific direction of the present embodiment refers to the horizontal direction (8), the vertical direction (7), the upper right direction (NE), and the upper left direction (Nff). When the image to be detected is blue (B), 'as in the fifth. In the figure A, the edge values in each direction can be obtained from the following formulas. To: v=[abs(Nc-dd+ABS(Nc_D6)]/2; Form No. A0101 帛9 pages/Total 26 pages 1002000107-0 201230792 H=[ABS(Nc-D3)+ABS(Nc~D4)] /2; NE=[ABS(Nc-D2)+ABS(Nc-D5)]/2; NW=[ABS(Nc-D0)+ABS(Nc-D7)]/2.

[0030] Next, according to the specific direction obtained in step 22, in step 232, it is determined whether the edge value of the direction is greater than a critical value. In this embodiment, it is determined whether half of the edge value edge_value is greater than the edge threshold B_edge_value_th. If the determination is YES, the pixel to be detected has non-smoothness. Steps 231-232 can be expressed as follows: if B(X,Y)=B_H then edge_value=H; else if B(X,Y)=B_V then edge_value=V; else if B(X,Y) = B_NE theri edge^ Value^NE; .....5 '..... else edge—value=NW; if edge_value/2>B-edge—value a th then Nc is non-smooth. The above step 23 232 is waiting The description is made when the detection pixel value Nc is blue (B), however, the same steps are equally applicable to the case where the pixel value to be detected is red (R), as shown in FIG. The difference is only in the replacement of B in each variable with R. The above step 231_232 can also be applied to the case where the pixel value Nc to be detected is green (G), as shown in the figure. The only difference is that the B in each variable is replaced by g. According to the flow shown in the first figure above, if the various decisions of steps 21 to 23 can be met, it can be determined that the pixel value Nc to be detected is a bad point. When it is judged to be a dead pixel, the pixel value of the dead pixel can be corrected or compensated (referred to as "complement point"). In this embodiment, the compensation of the dead pixels is performed based on the adjacent color pixel values of the dead pixels. 100100060 Form No. A0101 Page 10 of 26 1002000107-0 201230792 - [0032] Ο [0033]

[0035] The ninth diagram shows the fill point flow when the dead point Nc is blue (as shown in the fifth diagram) or red (as shown in FIG. 5B). First, in step 91, eight adjacent color-matching pixel values D0-D7 are sorted to obtain a maximum value. Next, in step 92, according to the maximum value, five adjacent color-matching pixel values far from the maximum value are weighted and averaged together with the pixel value Nc of the dead pixels to obtain a compensation value of the dead pixels. For example, when D0 is the maximum value, the values of five adjacent color pixels are D2, D5, D4, D6, and D7; when D1 is the maximum value, the values of five adjacent color pixels are D3, D4, and D5. D6, D7. In one embodiment, the weight of the dead pixels is 3/8, and the weights of the five adjacent color pixels are all 1/8. In another embodiment, the dead pixels have a weight of 1/16, and the five adjacent color-coded pixels have a weight of 3/16. The tenth graph shows the fill point flow when the dead point Nc is green. The eleventh figure shows the green (G) dead pixel value Nc and the adjacent color pixel values N0-N7. First, in step 101, the same color pixel values D0-D3 closest to the dead pixels Nc are sorted to obtain a sorting minimum value minl_g, which can be expressed as follows: minl_g=sorting(D0-D3). In step 102, the foregoing The same color pixel values D0-D3 closest to the dead point Nc are calculated to obtain the nearest pixel average value avg_g, which can be expressed as follows: avg_g=avg(D0-D3). In step 103, four nearest dead pixels Nc are obtained. The homochromatic pixel values D0-D3 together with the median of the dead pixels Nc, and the four average color pixel values D4-D7 closest to the bad point Nc, together with the mean median of the dead pixels Nc, are obtained. Next, the average median_g of the two average medians is obtained. 100100060 Form No. A0101 Page 11 of 26 1002000107-0 201230792 [0036] Finally, in step 104, the partial statistical value obtained according to the above step 10, 1〇3, is obtained to obtain the compensation value of the dead point. In the embodiment, the compensation value of the dead point is equal to the average value of the mean median_g of the average median and the minimum value minl_g of the order, that is, avg (media_g, minl_g). In another embodiment, the offset value of the dead pixel is equal to the average of both the nearest pixel average aVg_g and the sort minimum minl-g, i.e., avg(avg_g, minl_g). In yet another embodiment, the penalty value of the dead point is equal to the sorted minimum heart! Or equal to the average median value of median_g; or equal to the nearest pixel average of 3^^. The above description is only for the preferred embodiment of the present invention, and is not intended to limit the invention of the invention. Any equivalent changes or modifications made in the spirit of the invention disclosed herein should be included. Within the scope of the patent application. [Simple description of the diagram] Reward-picture display-image sensing H array of light-sensitive elements and color chopping array. The second figure shows a flow chart of a method for detecting an image sensing bad pixel point (_P i X e 1 ) according to an embodiment of the present invention. The second to second figures D show the pixel value to be detected and the pixel value of the adjacent position. The fourth figure shows a detailed flowchart of the step 2 1 of the first figure. The fifth to fifth C pictures show the pixel values to be detected and the adjacent color pixel values. The sixth and sixth panels show the horizontal direction enhancement weight, the vertical direction enhancement weight, the upper right direction enhancement weight, and the upper left direction enhancement weight. 100100060 Form No. A0101 Page 12 of 26 1002000107-0 201230792 The seventh diagram shows the detailed flow chart of step 2 2 of the second figure. The eighth figure shows a detailed flow chart of step 2 3 of the second figure. The ninth figure shows the fill point flow when the dead point is blue or red. The tenth figure shows the fill point flow when the dead point is green. The eleventh figure shows the green dead pixel value and the adjacent color pixel value. [Main component symbol description] [0039] Ο

10 Photosensor array 12 Color filter array 21-23 Steps 211-215 Steps 221-224 Steps 231-232 Steps 91-92 Steps 101-104 Step Nc Pending pixels/Bad points N0-N7 Pixels R Red G Green B Blue Color 100100060 Form number Α 0101 Page 13 / Total 26 pages 1002000107-0

Claims (1)

201230792 VII. Patent application scope: 1. A method for detecting image-detecting bad pixel points, including: Good view--a difference between the pixel value to be detected and the pixel value of the plurality of adjacent positions, the distance pixel value and the complex adjacent pixel value Continuity in the direction and (4)-specific direction; and ', according to the particular direction of the piece, the smoothness of the pixel value to be detected in the particular direction is examined. 2. The method for detecting a bad pixel in an image according to claim 1, wherein at least a portion of the adjacent pixel is different in color from the pixel to be detected. 3. The method for detecting an image of a bad pixel according to the first aspect of the patent application, wherein the pixel to be detected is used as a center of the circle, and forms a 3×3 matrix ij with the pixel of the adjacent position. 4. The method for detecting an image of a bad pixel according to the first aspect of the patent application, wherein the difference detection step comprises: obtaining a difference between an average value of a pixel value of the money to be detected; Obtaining, as a horizontal neighboring metric, a difference ' between the pixel value and the horizontal neighboring pixel value as a horizontal neighboring metric value; obtaining an average of the pixel value to be detected and the pixel value of the four corner neighboring positions a difference value of the value, as a corner neighboring metric value; according to the riding straight neighboring metric value, the horizontal neighboring metric value and the corner neighboring metric value 'to obtain - the integrated neighboring metric value; and - "T, the neighboring metric value and the Absolute difference of the pixel value to be detected 100100060 Form number A0101 1002000107-0 201230792 Ο 7 ο 8 9 100100060 Pixel point = 2 metric threshold value, if the determination is YES, then the value to be detected, 2%% The detection method of the image sensing bad pixel point is right. The average 'value of the pixel value to be detected and the vertical neighboring pixel value is less than G', so that the vertical neighboring metric value is 〇; The average value of the pixel value to be detected and the pixel value of the horizontal neighboring position is less than 0', so that the horizontal neighboring metric value is 〇; if the difference between the pixel value of the pixel to be detected and the pixel value of the adjacent position of the four corners is less than 0 , the neighboring metric value of the corner is 0; and the comprehensive neighboring metric value of the comprehensive person Zheng Ruojin 1 is greater than the pixel value to be detected, so that the σ X magnitude is equal to the pixel value to be detected β, such as = patent item first item The image sensing method for detecting a bad pixel point, the pixel to be detected is formed as a central pixel point to form a 3×3 array. The adjacent color image is as described in claim 1 of the patent application scope, wherein the image is detected.敎 (4) includes water in the direction of n direction and upper left direction. The detection method of the image sensing bad pixel point in the vertical direction and the upper right side = = item, wherein the above-mentioned continuity inspection step includes: using -増_重A method for detecting a bad pixel point of the image in which the pixel value 0 ===_8 is enhanced for a specific direction to be viewed, wherein the step of viewing the continuity includes::, after weighting a row of pixel values, taking an absolute Value For the parent _ and the value of the _ and the value of the money form number page 15 page, a total of 26 pages, the color pair and the value of 1002000107-201230792, obtain a horizontal continuum metric, a vertical direction of continuous metrics, a right upper direction The continuous metric value and the continuous metric value in the upper left direction obtain the continuous metric value in the horizontal direction, the continuous metric value in the vertical direction, the continuous metric value in the upper right direction, and the maximum or minimum value among the continuous metric values in the upper left direction. The value, as a comprehensive continuous metric; and determining whether the integrated continuous metric is greater than a continuous metric threshold, and if the determination is yes, the pixel to be detected has discontinuity. 10. The method for detecting an image of a bad pixel according to claim 1, wherein the step of detecting the smoothness comprises: obtaining an edge value of each of the specific directions; and determining an edge value of the specific direction. Whether it is greater than a critical value, if the determination is yes, the pixel to be detected has non-smoothness. 11. The method for detecting an image of a bad pixel according to claim 1, wherein the pixel value to be detected is different from the pixel value of the adjacent position, and the pixel value to be detected and the pixel value of the adjacent color are specific to The direction of the pixel to be detected is non-smooth, and the value of the pixel to be detected is non-smooth. 12. The method for detecting an image of a bad pixel according to claim 11 of the patent application, further comprising: performing a compensation of a dead pixel according to a plurality of pixels of the same point of the same color. 13. The method for detecting an image of a bad pixel according to claim 12, wherein when the dead point is blue or red, the step of complementing comprises: sorting the plurality of adjacent color pixels to obtain a Sorting the maximum value; and 100100060 Form number A0101 Page 16 of 26 Page 1002000107-0 201230792 14 . According to the sorting maximum value, the complex adjacent color pixel value farther from the sorting maximum value, together with the pixel of the dead pixel The values are weighted averaged to obtain a compensation value for the dead point. The method for detecting a bad pixel point of the image according to claim 12, wherein when the dead point is green, the step of complementing comprises: sorting the plurality of same-color pixel values closest to the bad point to obtain a Sorting the minimum value; calculating the same-color pixel value closest to the bad point to obtain 最近 a nearest pixel average value; obtaining a complex same-color pixel value closest to the bad point together with an average median value of the bad point (median) And obtaining a plurality of same-color pixel values that are next to the bad point together with an average median value of the bad points, and obtaining an average of the two average median values; and according to the sorted minimum value, the nearest pixel average value, and the second The portion of the average of the median values to obtain the compensation value for the dead point. ο 100100060 Form No. A0101 Page 17 of 26 1002000107-0