CN100584030C

CN100584030C - Lost color value component reconstruction method and device

Info

Publication number: CN100584030C
Application number: CN200610139642A
Authority: CN
Inventors: 彭起凤; 赖秋原
Original assignee: MStar Semiconductor Inc Taiwan
Current assignee: MediaTek Inc
Priority date: 2006-09-20
Filing date: 2006-09-20
Publication date: 2010-01-20
Anticipated expiration: 2026-09-20
Also published as: CN101150734A

Abstract

The invention discloses a lost color value component reconstruction method and a lost color value component reconstruction device, which can improve the definition of an edge image after color value reconstruction. The method is applied to the pixel points to be reconstructed in the Bayer pattern pixel point matrix, and comprises the following steps: calculating a vertical minimum color difference change estimated value Cv and a horizontal minimum color difference change estimated value Ch according to the acquired equal color value components on a first group of pixels surrounding the pixel to be reconstructed in the pixel matrix, wherein the first group of pixels comprises other pixels besides pixels in the same row or column as the pixel to be reconstructed; and determining the second color value component according to the vertical minimum color difference change estimation value Cv and the horizontal minimum color difference change estimation value Ch. In addition, a lost color value component reconstruction device is also provided.

Description

Lost color vector reconstruction method and device

Technical field

The present invention relates to a kind of lost color vector reconstruction method, finger is applied to the lost color vector reconstruction method for the treatment of on the reconstructed image vegetarian refreshments in the Bayer pattern pixel matrix especially.Also relate to a kind of lost color vector reconstruction device in addition.

Background technology

In camera market now, digital camera almost has been substituted traditional camera and has become main flow, wherein to be in traditional camera be to utilize photographic film to come sensing image to Zui Da difference, digital camera then is to utilize Charged Coupled Device (Charge Coupled Device is called for short CCD) or CMOS transducer (CMOS sensor) to come sensing image.But, because Charged Coupled Device or CMOS transducer can only sense the power of light, can not sense change in color.Therefore, when digital camera was wanted to have the ability of chromatic image sensing, the producer must add that colored filter (Color Filter is called for short CF) carries out the action of color separation in the front of Charged Coupled Device or CMOS transducer.

And colored filter all is to adopt RGB (RGB) three primary colors to carry out color separation usually, so need the pixel of three Charged Coupled Devices to capture respectively, a last pixel that again three color-values that pixel captured of three Charged Coupled Devices is mixed in the full-color image.But because the consideration on cost and the size, general digital camera only can use the pixel of single Charged Coupled Device to carry out the color acquisition, so will make each pixel can only capture a certain color-values component in RGB (RGB) three primary colors, and cause on the co-located other two color-values components to lose.

And be under low-cost and undersized restriction, to give for change because of saving the color-values component that Charged Coupled Device quantity is lost, following method for reconstructing just is developed.See also first figure, it is modal colorized optical filtering array (Color Filter Array in the present digital camera, be called for short CFA) the pattern schematic diagram, wherein each square corresponds to the pixel in the Charged Coupled Device, and by graphic know find out, green in the pattern (G), blue (B) are 2: 1: 1 with the pixel quantity ratio of red (R), and this kind arrangement mode is commonly called Bayer pattern (Bayerpattern), and its details can be with reference to No. 3971065 US Patent specification.So the color-values component that must utilize colorized optical filtering array figure shown in first figure and Charged Coupled Device to arrange in pairs or groups and capture carries out the mathematical operation of interpolative operation method again, use supposition and reconstruct other color-values that each pixel is lost.

And it is very many about the kind of interpolative operation method, but can be divided into two classes basically, the first kind is called fixed image interpolarting method, and the interpolating method that belongs to this type of has neighbor point (nearest) interpolation method, bilinearity (bilinear) interpolation method and color to change (Smooth Hue Transition) interpolation method or the like gently.Therefore but the class interpolation method is when carrying out the color-values that interpolative operation obtains to lose in a certain pixel, its neighbor point of getting is fixed with respect to the relative position of this pixel, so itself does not detect the ability at edge this type of interpolation method, so its image edge lines of rebuilding out partly can produce the phenomenon of image fog.

And be to improve above-mentioned disappearance, second class methods that are called free-standing image interpolarting method just are developed, this type of interpolation method carry out interpolative operation obtain in a certain pixel to lose fall color-values the time, its neighbor point of getting is unfixed with respect to the relative position of this pixel, meaning is the ability that this type of interpolation method itself has the detecting edge, therefore can reduce ill-defined phenomenon.For example No. 5629734 specification of United States Patent (USP) just described and utilized this type of interpolation method to carry out technological means and interlock circuit that color-values is rebuild, but the disappearance of this case technology is that it only utilizes this pixel of rebuild losing color-values with desire to go together and the color-values information of the neighborhood pixels point of same column is carried out the judgement and the selection of sampling point, cooperate shown in Figure 1, before this selected neighborhood pixels point of case all be positioned at desire rebuild to lose color-values pixel 1 the colleague and with listing (two arrow lines as shown in FIG.), so the image of rebuilding only can reduce ill-defined phenomenon limitedly.

Comprehensive above-mentioned reason can be learnt, present color-values reconstruction method still can't reflect the image edge of all directions effectively, therefore the color error after rebuilding still has sizable space of improving, and how to improve the disappearance that above-mentioned existing means go up in this respect, for developing topmost purpose of the present invention.

Summary of the invention

First technical problem to be solved by this invention provides a kind of lost color vector reconstruction method, and it can improve the definition of edge images whose chromatic after color-values is rebuild.

In addition, second technical problem to be solved by this invention provides a kind of lost color vector reconstruction device, uses the definition of edge images whose chromatic after this device can improve the color-values reconstruction.

In order to solve above first technical problem, the invention provides a kind of lost color vector reconstruction method, being applied to one in the Bayer pattern pixel matrix treats on the reconstructed image vegetarian refreshments, this treats that the reconstructed image vegetarian refreshments only has one first color-values component and loses one second color-values component and one the 3rd color-values component, this first color-values component in this pixel matrix wherein, the ratio of this second color-values component and the 3rd color-values component is 1: 2: 1, wherein, method for reconstructing comprises the following step: treat that according to surrounding this in this pixel matrix these color-values components that obtained on one first group of pixel of reconstructed image vegetarian refreshments calculate the minimum aberration of a vertical direction and change estimated value Cv and the minimum aberration variation of horizontal direction estimated value Ch, wherein in this first group of pixel except have treat with this that reconstructed image vegetarian refreshments is gone together or the pixel of same column also include other pixel; And change estimated value Cv according to the minimum aberration of this vertical direction and determine this second color-values component with the minimum aberration variation of horizontal direction estimated value Ch.

In order to solve above second technical problem, the invention provides a kind of lost color vector reconstruction device, being applied to one in the Bayer pattern pixel matrix treats on the reconstructed image vegetarian refreshments, this treats that the reconstructed image vegetarian refreshments only has one first color-values component and loses one second color-values component and one the 3rd color-values component, this first color-values component in this pixel matrix wherein, the ratio of this second color-values component and the 3rd color-values component is 1: 2: 1, wherein, and this reconstructing device comprises: a minimum aberration changes estimator, it is to read in this pixel matrix to surround this and treat that these color-values components that obtained on one first group of pixel of reconstructed image vegetarian refreshments calculate the minimum aberration of a vertical direction and change estimated value Cv and the minimum aberration variation of horizontal direction estimated value Ch, wherein in this first group of pixel except have treat with this that reconstructed image vegetarian refreshments is gone together or the pixel of same column also include other pixel; One compares determining device, signal is connected in this minimum aberration and changes estimator, it changes estimated value Cv for the minimum aberration of this vertical direction of reception and the minimum aberration of this horizontal direction changes estimated value Ch, and a Cv-Ch and a threshold value T are compared, to produce a comparison signal; And a color value arithmetic device, signal is connected in this comparison determining device, in order to produce this second color-values component in response to this comparison signal.

Because the present invention not only utilizes and treats that the reconstructed image vegetarian refreshments is gone together and the color-values information of the neighborhood pixels point of same column is carried out the judgement and the selection of sampling point, but also with reference to the color-values information of other area pixel point, so the edge blurry of reconstructed image can significantly reduce, therefore can effectively improve the disappearance of prior art.

Description of drawings

Below in conjunction with the drawings and specific embodiments the present invention is described in further detail.

Fig. 1 is the schematic diagram of modal colorized optical filtering array patterns in the present digital camera;

Fig. 2 is to be the Bayer pattern schematic diagram of 5 * 5 pixels at center with red pixel point;

Fig. 3 is to be the Bayer pattern schematic diagram of 5 * 5 pixels at center with blue pixel point;

Fig. 4 is to be the Bayer pattern schematic diagram of 5 * 5 pixels at center with green pixel point;

Fig. 5 is the steps flow chart schematic diagram that estimates out green component G13 in this case lost color vector reconstruction method;

Fig. 6 is about the function block schematic diagram of lost color vector reconstruction device in the specific embodiments of the invention.

The primary clustering symbol description.

Desire is rebuild the pixel 1 of losing color-values

Minimum aberration changes relatively determining device 41 of estimator 40

Color-values arithmetic unit 42 memory devices 43

Embodiment

See also Fig. 2, it for the purpose of following explanation is convenient, all compiles number P1～P25 with each pixel for being Bayer pattern (Bayer pattern) schematic diagram of 5 * 5 pixels at center with red pixel point.

Because center point P 13 is in actual color acquisition process, has only the color-values R13 that captures red component, and and fail to capture color-values G13, B13 green and blue component, therefore this case G13 and B13 of providing following method to reconstruct P13, and be convenience on describing, how explanation G13 rebuilds earlier.And in general situation, during loss color-values component on rebuilding the P13 point, the loss color-values component of P1 to P12 has been rebuild all and has been finished.

And main spirit of the present invention is to utilize the close theory of an aberration to calculate, its main idea is in this zone that the difference of two color-values components on certain a bit can equal the mean value of two color-values component differences on each point in this zone or the consecutive points, G-R=G '-R ' for example, G-B=G '-B ', wherein R, B, G represent the color-values component of this point, and G '-R ', G '-B ' then represents the mean value of green red and turquoise color-values component difference on each point in this zone or the consecutive points.

So use the close theory of above-mentioned aberration, define following equation:

Gv-R13＝(k1*((G8-R3)+(G18-R23))+k2*((G8-R13)+(G18-R13)))/k3；

Gh-R13＝(k1*((G12-R11)+(G14-R15))+k2*((G12-R13)+(G14-R13)))/k3；

Wherein to be defined as respectively with P13 be the green estimated value of vertical direction and the green estimated value of horizontal direction at center for Gv and Gh; Equation the right then is the mean value of green-red color-values component difference on the mean value of green-red color-values component difference and the horizontal direction consecutive points on the vertical direction consecutive points in the Bayer pattern (Bayer pattern) of 5 * 5 pixels, wherein k1, k2 and k3 are default coefficient, and R13 represents the red color value component of P13, the rest may be inferred, G8 represents the green tint value component of P8, then represents red color value component of P3 or the like as for R3.

Convenient for describing, with above-mentioned be that the mean value of green-red color-values component difference on the mean value of green-red color-values component difference on the vertical direction consecutive points at center and the horizontal direction consecutive points is represented with GRv and GRh respectively with P13, wherein

GRv＝Gv-R13＝(k1*((G8-R3)+(G18-R23))+k2*((G8-R13)+(G18-R13)))/k3；

GRh＝Gh-R13＝(k1*((G12-R11)+(G14-R15))+k2*((G12-R13)+(G14-R13)))/k3。

Then, utilize the following equation of the close theoretical definition of same aberration again:

Bv-Gv＝(k4*(B7-G12)+k5*(B17-G12)+k6*(B9-G14)+k7*(B19-G14))/k8；

Bh-Gh＝(k4*(B7-G8)+k5*(B17-G18)+k6*(B9-G8)+k7*(B19-G18))/k8；

Wherein k4, k5, k6, k7 and k8 are default coefficient, and B7 represents the blue color-values component of P7, and the rest may be inferred, and G12 represents green tint value component of P12 or the like.

In this embodiment, because R13 is known, so Gv and Gh can try to achieve respectively with GRv+R13 and GRh+R13, therefore is that the blue estimated value Bv of vertical direction and the blue estimated value Bh of horizontal direction at center also can be estimated out respectively with P13:

Bv＝Gv+(k4*(B7-G12)+k5*(B17-G12)+k6*(B9-G14)+k7*(B19-G14))/k8；

Bh＝Gh+(k4*(B7-G8)+k5*(B17-G18)+k6*(B9-G8)+k7*(B19-G18))/k8；

And be can describe convenient, with above-mentioned be that the mean value of vertical direction consecutive points blueing-green tint value component difference at center and the mean value of horizontal direction consecutive points blueing-green tint value component difference are represented with BGv and BGh respectively with P13;

BGv＝((k4*(B7-G12)+k5*(B17-G12)+k6*(B9-G14)+k7*(B19-G14))/k8)；

BGh=((k4* (B7-G8)+k5* (B17-G18)+k6* (B9-G8)+k7* (B19-G18))/k8); Conversely then

GBv＝-((k4*(B7-G12)+k5*(B17-G12)+k6*(B9-G14)+k7*(B19-G14))/k8)；

GBh＝-((k4*(B7-G8)+k5*(B17-G18)+k6*(B9-G8)+k7*(B19-G18))/k8)，

Utilize same theory to list following two formulas again and define green-red variation value GRavg and green-blue variation value GBavg on the whole on the whole in this pixel matrix;

GRavg＝Gavg-Ravg

GBavg＝Gavg-Bavg

And wherein Gavg calculates gained, the following formula of definable from the green tint value component that with P13 is at least 18 pixels 5 * 5 pixels at center (be not limited to known or forecasting institute gets):

Gavg=(a1* (G8+G12+G14+G18)+a2* (G2+G4+G6+G10+G16+G20+G22+G24)+a3* (G1+G3+G5+G7+G9+G11))/a4; Wherein a1, a2, a3 and a4 are default coefficient.

Ravg calculates gained, the following formula of definable from the red color value component that with P13 is at least 14 pixels 5 * 5 pixels at center (be not limited to known or forecasting institute gets):

Ravg=(a5*R13+a6* (R1+R5+R21+R25)+a7* (R3+R11+R15+R23)+a8* (R2+R6)+a9* (R4+R8+R12))/a10; Wherein a5, a6, a7, a8, a9 and a10 are default coefficient.

As for Bavg then is to calculate gained, the following formula of definable from the blue color-values component that is at least 12 pixels 5 * 5 pixels at center (be not limited to known or forecasting institute gets) with P13:

Bavg=(a11* (B7+B9+B17+B19)+a12* (B2+B4+B6+B10)+a13* (B3+B8)+a14* (B1+B5))/a15; Wherein a11, a12, a13, a14 and a15 are default coefficient.

And after obtaining above-mentioned data, it is the minimum aberration variation estimated value Cv of vertical direction and the minimum aberration variation of the horizontal direction estimated value Ch at center that present embodiment utilizes following formula to obtain with P13 again, wherein

Cv＝e1*|GRv-GRavg|+e2*|GBv-GBavg|+e3*|R3-R13|+e4*|R23-R13|+e5*|G18-G8|+e6*|G7-G12|+e7*|G9-G14|

Ch=e1*|GRh-GRavg|+e2*|GBh-GBavg|+e3*|R11-R13|+e4*|R15-R1 3|+e5*|G14-G12|+e6*|G7-G8|+e7*|G9-G8|; Wherein e1, e2, e3, e4, e5, e6 and e7 are default coefficient.

It is the center that minimum aberration variation estimated value Cv of vertical direction and the minimum aberration variation of horizontal direction estimated value Ch represent with this pixel P13 respectively, respectively vertically with the color change degree of horizontal direction, the color change degree of big more this direction of representative of estimated value is big more, otherwise, then represent the color change degree of this direction smaller when estimated value is more little.

In view of the above, in present embodiment, this pixel P13 goes up the green tint value component G13 of leakage originally, and available following formula pushes away:

(1) if Cv-Ch＜T represents vertical direction color change degree smaller, then G13=first estimated value=Gvs.

(2) if Cv-Ch＞T represents horizontal direction color change degree smaller, then G13=second estimated value=Ghs.

(3) if Cv-Ch=T represents vertical direction identical with horizontal direction color change degree, G13=the 3rd estimated value=(G8+G12+G18+G14)/4 then.

Wherein T is a predetermined threshold value, and the first estimated value Gvs and the second estimated value Ghs then push away with following formula:

Gvs＝s1*R13+s2*(s3*((G8-R3)+(G18-R23))+s4*((G8-R13)+(G18-R13)))/s5+s6*Bv+s7*GBavg；

Wherein:

s2*(s3*((G8-R3)+(G18-R23))+s4*((G8-R13)+(G18-R13)))/s5

Representative treats that with this reconstructed image vegetarian refreshments P13 is center and the mean value of representing in this pixel matrix this second color-values component and this first color-values component difference on the vertical direction consecutive points.

Ghs＝s1*R13+s2*(s3*((G12-R11)+(G14-R15))+s4*((G12-R13)+(G14-R13)))/s5+s6*Bh+s7*GBavg；

Wherein:

s2*(s3*((G12-R11)+(G14-R15))+s4*((G12-R13)+(G14-R13)))/s5

Representative treats that with this reconstructed image vegetarian refreshments P13 is center and the mean value of representing in this pixel matrix this second color-values component and this first color-values component difference on the horizontal direction consecutive points, is default coefficient as for s1, s2, s3, s4, s5, s6 and s7.

After obtaining G13, pixel P13 goes up the blue color-values component B13 of leakage originally, then can utilize with P13 is that at least 17 color-values components in 5 * 5 pixels at center (be not limited to known or forecasting institute gets) calculate gained, and its algorithm can be as following formula:

B13=G13+ (f1* ((B7-G7)+(B9-G9)+(B17-G17)+(B19-G19))+f2* ((B8-G8)+(B14-G14)+(B18-G18)+(B12-G12)))/f3; Wherein f1, f2 and f3 are default coefficient.

In sum, be the Bayer pattern (Bayerpattern) of 5 * 5 pixels at center with red pixel point R13, can utilize said method to obtain the G13 and the B13 of loss.In like manner as can be known, as shown in Figure 3 be the Bayer pattern (Bayerpattern) of 5 * 5 pixels at center with blue pixel point B13, because blue pixel point B13 and red pixel point R13 are all and respectively account for 25 percent ratio, also can utilize above-mentioned G13 and the R13 that obtains loss with quadrat method.

As for shown in Figure 4 be the Bayer pattern (Bayer pattern) of 5 * 5 pixels at center with green pixel point G13, it is that the color-values component (be not limited to known or forecasting institute gets) of at least 9 pixels in 5 * 5 pixels at center calculates gained that the R13 of its loss and B13 then can utilize with G13, and its algorithm can be as following formula:

R13＝G13+(w1*((R12-G12)+(R14-G14))+w2*((R8-G8)+(R18-G18)))/w3

B13＝G13+(w1*((B8-G8)+(B18-G18))+w2*((B12-G12)+(B14-G14)))/w3；

Wherein w1, w2 and w3 are default coefficient.

And above-mentioned coefficient: k1～k8, a1～a15, e1～e7, s1～s7, f1～f3 and the w1～w3 of respectively organizing, it can utilize several reference patterns to carry out emulation (simulation) and calculate, come computing to approach out one group of preferable coefficient to minimize root-mean-square error, minimize root-mean-square error and can be the personage who knows this skill and understand, do not repeat them here.

See also Fig. 5 in addition again, it is for estimating out the steps flow chart schematic diagram of green component G13 in the above-mentioned lost color vector reconstruction method, at first, treat that according to surrounding this in this pixel matrix these color-values components that obtained on first group of pixel of reconstructed image vegetarian refreshments P13 calculate the minimum aberration of vertical direction and change estimated value Cv and the minimum aberration variation of horizontal direction estimated value Ch, wherein treat with this that reconstructed image vegetarian refreshments is gone together or the pixel of same column (P3 for example except having in this first group of pixel, P23, P8, P18, P11, P15 etc.) outer other pixel (P7 that also includes, P12, P9, P14, P8 etc.); Utilize minimum aberration variation estimated value Cv of vertical direction and the minimum aberration of horizontal direction to change estimated value Ch then and carry out following judgement:

(1) if Cv-Ch＜T represents vertical direction color change degree smaller, then G13=first estimated value=Gvs;

(2) if Cv-Ch＞T represents horizontal direction color change degree smaller, then G13=second estimated value=Ghs; And

Wherein this first estimated value Gvs treats that according to surrounding this in this pixel matrix these color-values components and the close theory of aberration that have obtained on one second group of pixel of reconstructed image vegetarian refreshments calculate, and wherein treats also to include other pixel (P7, P12, P9, P14, P17, P19 etc.) the pixel (for example P3, P23, P8, P18, P13 etc.) that the reconstructed image vegetarian refreshments goes together except having with this in this second group of pixel.

And this second estimated value Ghs treats that according to surrounding this in this pixel matrix these color-values components and the close theory of aberration that have obtained on one the 3rd group of pixel of reconstructed image vegetarian refreshments calculate, and wherein treats also to include other pixel (for example P7, P8, P9, P17, P18, P19 etc.) the pixel (for example P11, P12, P13, P14, P15 etc.) of reconstructed image vegetarian refreshments same column except having with this in the 3rd group of pixel.

Include in this pixel matrix around this as for the 3rd estimated value and to treat green tint value component average on one the 4th group of pixel (for example P8, P12, P18, P14) of reconstructed image vegetarian refreshments.

Fig. 6 is about the function block schematic diagram of lost color vector reconstruction device in the specific embodiments of the invention, it consists predominantly of minimum aberration and changes estimator 40, compare determining device 41, color-values arithmetic unit 42 and memory device 43 (memory device is memory device just), and wherein these memory device 43 signals are connected in this minimum aberration variation estimator 40 and this color-values arithmetic unit 42, mainly be Bayer pattern (Bayerpattern) pixel matrix, and provide this minimum aberration variation estimator and this color-values arithmetic unit to read and write in order to 5 * 5 pixels of storage shown in second figure.And minimum aberration changes estimator 40 and reads and surround this in this stored pixel matrix of memory device 43 and treat that these color-values components that obtained on this first group of pixel of reconstructed image vegetarian refreshments calculate the minimum aberration of this vertical direction and change estimated value Cv and the minimum aberration variation of this horizontal direction estimated value Ch.And signal is connected in the comparison determining device 41 that this minimum aberration changes estimator 40, it is to receive the minimum aberration of this vertical direction to change estimated value Cv and the minimum aberration variation of this horizontal direction estimated value Ch, and a Cv-Ch and a threshold value T compared, when Cv-Ch sends one first signal during less than this threshold value T, when Cv-Ch sends a secondary signal during greater than this threshold value T; And when equaling this threshold value T, sends Cv-Ch one the 3rd signal.At last, signal is connected in this comparison determining device 41 and the color-values arithmetic unit 42 that reads memory device 43, it is to read in response to the triggering of this first signal in the memory device 43 to surround this in this pixel matrix and treat that these color-values components that obtained on one second group of pixel of reconstructed image vegetarian refreshments and the close theory of aberration calculate one first estimated value and come to treat this second color-values component of reconstructed image vegetarian refreshments and write in the memory device 43 as this, read in response to the triggering of this secondary signal in addition in the memory device 43 and to surround this in this pixel matrix and treat that these color-values components that obtained on one the 3rd group of pixel of reconstructed image vegetarian refreshments and the close theory of aberration calculate one second estimated value and treat this second color-values component of reconstructed image vegetarian refreshments and write memory device 43 as this, wherein treat also to include other pixel the pixel that the reconstructed image vegetarian refreshments goes together except having in this second group of pixel, and treat also to include other pixel the pixel of reconstructed image vegetarian refreshments same column except having in the 3rd group of pixel with this with this.In sum, this case technology is not only utilized and is treated that the reconstructed image vegetarian refreshments is gone together and the color-values information of the neighborhood pixels point of same column is carried out the judgement and the selection of sampling point, but also with reference to the color-values information of other area pixel point, so the edge blurry of reconstructed image more can significantly reduce, therefore effectively improve existing means disappearance, and then reach development main purpose of the present invention; All other do not break away from the equivalence of being finished under the disclosed spirit and changes or modification, all should be included in the claim.

Claims

1. A method for reconstructing lost color value components, applied to a pixel to be reconstructed in a Bell pattern pixel matrix, the pixel to be reconstructed only has a first color value component and loses a second color value component and A third color value component, wherein the ratio of the first color value component, the second color value component and the third color value component in the pixel matrix is 1:2:1, characterized in that the reconstruction method includes the following step:

According to the obtained color value components of a first group of pixels surrounding the pixel to be reconstructed in the pixel matrix, an estimated minimum color difference change in the vertical direction Cv and an estimated minimum color difference change in the horizontal direction Ch are calculated, wherein The first group of pixels includes other pixels in addition to the pixels in the same row or column as the pixel to be reconstructed; and

The second color value component is determined according to the estimated minimum color difference change in the vertical direction Cv and the estimated minimum color difference change in the horizontal direction Ch.

2. The method for reconstructing lost color value components according to claim 1, wherein the step of determining the second color value component according to the estimated value of the minimum color difference change in the vertical direction Cv and the estimated value of the minimum color difference change in the horizontal direction Ch includes : When Cv-Ch is less than a threshold value T, the second color value component of the pixel to be reconstructed is a first estimated value, and the first estimated value is based on the pixels surrounding the pixel to be reconstructed in the pixel matrix The obtained color value components and color difference on a second group of pixels are calculated by similar theory, wherein the second group of pixels includes other pixels besides the pixel to be reconstructed.

3. The method for reconstructing lost color value components according to claim 1, wherein the step of determining the second color value component according to the estimated minimum color difference change in the vertical direction Cv and the estimated minimum color difference change in the horizontal direction Ch includes : When Cv-Ch is greater than the threshold value T, the second color value component of the pixel to be reconstructed is a second estimated value, and the second estimated value is based on the pixel matrix surrounding the pixel to be reconstructed The obtained color value components and color difference of the third group of pixels are calculated by similar theory, wherein the third group of pixels includes other pixels besides the pixel in the same column as the pixel to be reconstructed.

4. The method for reconstructing lost color value components as claimed in claim 1, wherein Cv has components of the following values:

Taking the pixel to be reconstructed as the center and representing the average value of the difference between the second color value component and the first color value component at adjacent points in the vertical direction of the pixel matrix, and the pixel to be reconstructed as the center and representing The absolute value of the average value of the difference between the second color value component and the first color value component in the pixel matrix as a whole;

Taking the pixel to be reconstructed as the center and representing the average value of the difference between the second color value component and the third color value component at adjacent points in the vertical direction of the pixel matrix, and the pixel to be reconstructed as the center and representing The absolute value of the average value of the difference between the second color value component and the third color value component in the pixel matrix as a whole;

Taking the pixel to be reconstructed as the center and representing the difference of the first color value component between two pixels on the vertical axis in the pixel matrix; and

Taking the pixel to be reconstructed as the center and representing the difference of the second color value component between two pixels in the vertical direction on both sides of the vertical axis in the pixel matrix.

5. The method for reconstructing lost color value components according to claim 1, wherein the estimated value Ch of the minimum color difference change in the horizontal direction has components of the following values:

Taking the pixel to be reconstructed as the center and representing the average value of the difference between the second color value component and the first color value component at adjacent points in the horizontal direction of the pixel matrix, and the pixel to be reconstructed as the center and representing The absolute value of the average value of the difference between the second color value component and the first color value component in the pixel matrix as a whole;

Taking the pixel to be reconstructed as the center and representing the average value of the difference between the second color value component and the third color value component at adjacent points in the pixel matrix in the horizontal direction, and the pixel to be reconstructed as the center and representing a component of an absolute value between the mean values of differences between the second color value component and the third color value component as a whole in the pixel matrix;

Taking the pixel to be reconstructed as the center and representing the difference of the first color value component between two pixels on the horizontal axis in the pixel matrix; and

Taking the pixel to be reconstructed as the center and representing the difference of the second color value component between two pixels in the horizontal direction on both sides of the horizontal axis in the pixel matrix.

6. The method for reconstructing lost color value components according to claim 2, wherein the first estimated value includes the first color value component of the pixel to be reconstructed, and the pixel to be reconstructed as the center And it represents the average value of the difference between the second color value component and the first color value component on adjacent points in the vertical direction in the pixel matrix, centering on the pixel to be reconstructed and representing the vertical direction in the pixel matrix The estimated value of the third color value component is a component centered on the pixel to be reconstructed and representing the average value of the difference between the second color value component and the third color value component in the pixel matrix as a whole.

7. The method for reconstructing lost color value components according to claim 3, wherein the second estimated value includes the first color value component of the pixel to be reconstructed, and the pixel to be reconstructed as the center And it represents the average value of the difference between the second color value component and the first color value component on adjacent points in the horizontal direction in the pixel matrix, centering on the pixel to be reconstructed and representing the horizontal direction in the pixel matrix The estimated value of the third color value component and a component centered on the pixel to be reconstructed and representing the average value of the difference between the second color value component and the third color value component in the pixel matrix as a whole.

8. The method for reconstructing lost color value components according to claim 1, wherein the step of determining the second color value component according to the estimated minimum color difference change in the vertical direction Cv and the estimated minimum color difference change in the horizontal direction Ch includes : When Cv-Ch is equal to the threshold value T, the second color value component of the pixel to be reconstructed is a third estimated value, and the third estimated value includes the pixels surrounding the pixel to be reconstructed in the pixel matrix An average of the second color value components on the fourth group of pixels.

9. The method for reconstructing lost color value components according to claim 1, further comprising the following step: after obtaining the second color value component of the pixel to be reconstructed, using the pixel to be reconstructed The second color value component and the average of the difference between the third color value component and the second color value component on the eight pixels surrounding the pixel to be reconstructed are calculated, and then the pixel to be reconstructed is obtained The third color value component.

10. A lost color value component reconstruction device, applied to a pixel to be reconstructed in a Bell pattern pixel matrix, the pixel to be reconstructed only has a first color value component and loses a second color value component and A third color value component, wherein the ratio of the first color value component, the second color value component and the third color value component in the pixel matrix is 1:2:1, characterized in that the reconstruction device includes :

A minimum color difference change estimator, which reads the obtained color value components on a first group of pixels surrounding the pixel to be reconstructed in the pixel matrix to calculate a vertical minimum color difference change estimate value Cv and a The estimated value Ch of the minimum color difference change in the horizontal direction, wherein the first group of pixels includes other pixels in addition to the pixels in the same row or column as the pixel to be reconstructed;

A comparison judge, the signal is connected to the minimum color difference change estimator, which is to receive the minimum color difference change estimate Cv in the vertical direction and the minimum color difference change estimate Ch in the horizontal direction, and compare Cv-Ch with a threshold value T , to generate a comparison signal; and

A color value calculator, signally connected to the comparison judger, used for generating the second color value component from the comparison signal.

11. The device for reconstructing lost color value components as claimed in claim 10, wherein the comparison determiner sends out a first signal of the comparison signal when Cv-Ch is less than the threshold value T, and when Cv-Ch When it is greater than the threshold value T, a second signal of the comparison signal is sent, and the color value calculator is triggered by the first signal and according to a second group of pixels surrounding the pixel to be reconstructed in the pixel matrix Calculate a first estimated value as the second color value component of the pixel to be reconstructed based on the obtained color value components and color difference approximation theory, and trigger by the second signal according to the pixel matrix surrounded by A second estimated value is calculated as the second color value component of the pixel to be reconstructed by calculating the color value components obtained on a third group of pixels of the pixel to be reconstructed and the color difference approximation theory, wherein the first color value component of the pixel to be reconstructed is The second group of pixels contains other pixels in addition to the pixels in the same row as the pixel to be reconstructed, and the third group of pixels contains other pixels in addition to the pixels in the same column as the pixel to be reconstructed pixel.