KR101637488B1 - Image interpolation method and apparatus using pattern characteristics of color filter array - Google Patents

Abstract

The present invention relates to an image interpolation method and apparatus, and an image interpolation method according to an embodiment of the present invention includes calculating a horizontal / vertical absolute channel difference from a CFA image, estimating DAIDs of R, G, By interpolating along the horizontal / vertical direction, the peak signal-to-noise ratio (PSNR) and the structural similarity (SSIM) are large, and the visual quality is also excellent.

Adaptive, CFA, weighting, di-hatting, interpolation

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image interpolation method and apparatus using pattern characteristics of a color filter array,

The present invention relates to image processing, and more particularly, to an image interpolation method and apparatus.

Most digital cameras and camcorders capture color images through color CCD or CMOS. The acquired image is a hatched image according to a pattern of a color filter array (hereinafter, referred to as 'CFA'), and a hatched image includes two unknown color values for each pixel and one known color Value. The CFA pattern consists of the red (R), green (G) and blue (B) channels, and each channel is decimated by another factor. The decimated G channel has twice as many pixels as the R / B channel. The R / B colors alternate with the G color along the rows or columns. Alternating between G and R or G and B.

Demosaicing, or CFA interpolation, is a process of reconstructing unknown R, G, and B components to produce a full-color image. The two unknown colors of each pixel are estimated by various demosaicing algorithms. The de-mosaicing algorithm is to reconstruct the full-color image from incomplete data, i.e., two missing data at each pixel. This reconfiguration algorithm uses interchannel and intrachannel correlation of the data available on the RGB channels.

Conventional adaptive demosaicing algorithms consider color correlation and local characteristics of images. Edge-directed interpolation and various adaptive weighted sum interpolations use adaptive weights for interpolation of unknown pixel values. In addition, in the high frequency region where artifacts such as zipper effect, false color, and aliasing appear, the color difference of the channels, that is, the color difference between the R channel and the G channel, the B channel and the G channel Lt; / RTI > have high correlation so that conventional demagnification algorithms use these properties.

A conventional adaptive demagnification algorithm is described in X. Li, "Demosaicing by successive approximation," IEEE Trans. Image Process., Vol. 14, no. 3, pp. 370-379, Mar. 2005, and W. Lu et al., "Color filter array demosaicing: New method and performance measures," IEEE Trans. Image Process., Vol. 12, no. 10, pp. 1194-1210, Oct. 2003.

Li obtains the color difference after initial color interpolation of each color channel and finds the direction of the edge through the value to which the Laplacian filter is applied in the color difference domain to update the initial interpolated pixel values. Where the update is repeated until the stop criterion is satisfied to adjust the estimate of the unknown color pixel values by emphasizing the color difference rule in all iterations to suppress false colors and zip defects.

Lu is an edge sensing algorithm that uses a color difference model or a color-ratio model and preserves edges by interpolating along the edges while preventing interpolation across the edges. To estimate unknown color pixel values, these methods consider the spatial correlation between neighboring pixels and use the neighboring pixels to select the appropriate interpolation direction.

The prior art techniques described above produce high quality, demultiplexed images, and are particularly effective in reconstructing high frequency regions such as sharp edges. However, because Li repeats until the stop criterion is satisfied for updating the initial interpolated pixel, the computation speed is slow, and distortion occurs when the initial interpolated value is interpolated so that the edge direction is erroneous.

In addition, the prior art techniques described above are not sufficient to reduce artifacts such as false colors, aliasing, and zipper effects. Incorrect color around the edge and a detailed texture of the di-hatched image degrade the image quality. Overshoot defects, such as the zipper effect and the sharp color edge of the di-hatched image, are often another disadvantage of adaptive algorithms.

SUMMARY OF THE INVENTION The present invention has been made in order to solve the problems of the conventional techniques described above, and it is an object of the present invention to provide a method of detecting an edge direction from DAID (Differential of Absolute Inter- channel Differences) It is an object of the present invention to provide an image interpolation method and apparatus capable of interpolating three color pixel values.

According to an aspect of the present invention, there is provided an image interpolation method using characteristics of a color filter array pattern according to an exemplary embodiment of the present invention, Detecting edge directions by calculating absolute difference values between the left, right, up and down directions; Calculating directivity weights in the left, right, up and down directions according to pattern characteristics of the R, G, and B pixels, respectively; And interpolating an unknown pixel using the color difference values in the left, right, up and down directions from the R, G and B pixels according to the detected edge direction and the calculated direction weight, The absolute channel difference value is obtained by subtracting the absolute value of the difference between the color intensity value of the first channel and the color intensity value of the neighboring pixel of the second color channel on one side of the center pixel, Is calculated by subtracting the absolute value of the difference between the color intensity value and the color intensity value of the neighboring pixel of the second color channel.

Preferably, the absolute channel difference values are difference values of neighboring pixel values in the left, right, up, and down directions.

Preferably, the detecting step includes a step of detecting a horizontal difference (D _H ), which is a difference between the left and right direction absolute difference values in the left, right, upper and lower directions, And a difference (D _V ), respectively.

Preferably, in the detecting step, when the center pixel of the color filter array pattern is R _{i, j} , the horizontal difference D _H and the vertical difference D _V are calculated using the following equations (4) and .

Preferably, in the interpolation step, when the center pixel of the color filter array pattern is R _{i, j} , the unknown pixel G is interpolated using the following equation (6).

Preferably, the interpolation step interpolates G pixels among unknown pixels, and then interpolates the remaining unknown pixel values based on the interpolated G pixel values G '.

Preferably, in the interpolation step, when the center pixel of the color filter array pattern is B _{i, j} , the unknown R pixel is interpolated using the following equation (15).

,

,

,

인 것을 특징으로 한다.Preferably, when the center pixel of the color filter array pattern is G _{i, j} , the color difference values in the left, right, up, and down directions in Equation (15)

,

,

,

.

Advantageously, the image interpolation method further comprises adjusting the interpolated pixel value by adding the high frequency component value of the known pixel to the interpolated pixel value.

According to another aspect of the present invention, there is provided an image interpolating apparatus using characteristics of a color filter array pattern according to another embodiment of the present invention, An edge direction detecting unit for detecting an edge direction by calculating absolute difference values between left, right, up and down directions; A weight calculation unit for calculating directional weights in left, right, up and down directions according to pattern characteristics of the R, G, and B pixels; And an interpolator interpolating an unknown pixel using the color difference values in the left, right, up and down directions from the R, G and B pixels according to the detected edge direction and the calculated direction weight, Wherein the absolute channel difference value is a difference between a color intensity value of the first channel and a color intensity value of a neighboring pixel of the second color channel on one side of the center pixel, Is calculated by subtracting the absolute value of the difference between the color intensity value of the first color channel and the color intensity value of the neighboring pixel of the second color channel.

Preferably, the absolute channel difference values are difference values of neighboring pixel values in the left, right, up, and down directions.

Preferably, the edge direction detecting unit detects a horizontal difference (D _H ), which is a difference between the left and right direction absolute difference values in the left, right, upper and lower directions, and a difference value And a vertical difference (D _V ) that is a difference between the vertical difference and the vertical difference.

Preferably, the edge direction detecting unit calculates the horizontal difference (D _H ) and the vertical difference (D _V ) using the following equations (4) and (5) when the center pixel of the color filter array pattern is R _{i, j} .

Preferably, when the center pixel of the color filter array pattern is R _{i, j} , the interpolator interpolates the unknown pixel G using the following equation (6).

Preferably, the interpolator interpolates the G pixel among the unknown pixels, and then interpolates the remaining unknown pixel values based on the interpolated G pixel value (G ').

Preferably, when the central pixel of the color filter array pattern is B _{i, j} , the interpolator interpolates the unknown R pixel using the following equation (15).

,

,

,

인 것을 특징으로 한다.Preferably, when the center pixel of the color filter array pattern is G _{i, j} , the color difference values in the left, right, up, and down directions in Equation (15)

,

,

,

.

Preferably, the image interpolating apparatus using the characteristics of the color filter array pattern further comprises an adjustment unit for adjusting the interpolated pixel value by adding the high-frequency component value of the known pixel to the interpolated pixel value.

According to another aspect of the present invention, there is provided a recording medium having recorded thereon a program for causing a computer to execute the image interpolation method according to another embodiment of the present invention.

The proposed algorithm reduces vivid color edges and zipper effect around the texture. To evaluate the performance of the proposed algorithm, a conventional demagnification algorithm is simulated. Experimental results are compared in terms of peak signal-to-noise ratio (PSNR), structural similarity (SSIM) and visual quality. The performance of the proposed algorithm shows better performance than the known techniques and well reduces the zipper effect of color edges and textures.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description, only parts necessary for understanding the operation according to the present invention will be described, and descriptions of other parts may be omitted so as not to disturb the gist of the present invention.

In addition, terms and words used in the following description and claims should not be construed to be limited to ordinary or dictionary meanings, but are to be construed in a manner consistent with the technical idea of the present invention As well as the concept.

Figure 1 shows a Bayer CFA pattern, where half of the total number of pixels is green and the remaining pixels are equally assigned to red or blue. Each pixel in this pattern is only sensitive to one color, R, G, or B. Thus, color images obtained using this pattern are interpolated on three color channels to produce a full-color image, which is referred to as CFA demosaicing or interpolation.

The basic idea of a general adaptive weighting algorithm is to estimate the local variance from the hatched image and then use local variance for the demosaicing. Most conventional algorithms obtain a hatched image using the weighted sum of the intensity of neighboring pixels.

Figures 1a, 1b and 1c show a 5x5 Bayer CFA pattern, each having R, G and B pixels at the center of the pattern. Referring to the CFA pattern of FIG. 1A, an unknown G pixel is estimated first by considering the directionality of interpolation using horizontal and vertical gradients, as shown in the following equations (1) and (2).

이 다음 수학식 3과 같이 계산된다.Here, G _{p, q} and R _{p, q} are the known G and R pixel intensity values at (p, q) of the CFA pattern and are calculated using the horizontal gradient H _{i, j} and the vertical gradient V _i, G pixel value of

Is calculated as shown in the following equation (3).

는 상기 수학식 1 내지 3에서, R_p,q 를 B_p,q 로 대체하고 동일하게 계산하여 추정된다. 1B is similar to FIG. 1A except that R _{p, q} is replaced by B _{p, q} . Therefore, by using the CFA pattern of FIG. 1B, an unknown G pixel value,

Is estimated by substituting B _{p, q} with R _{p, q} in the above Equations 1 to 3 and calculating the same.

을 이용해서 미지의 R 픽셀 값 및 B 픽셀 값이 추정된다.1C shows a Bayer CFA pattern at a G-center pixel having an R pixel value and an B pixel value, which are unknown. Here, it is assumed that the high frequency components have similarity over the three color components, R, G and B,

An unknown R pixel value and a B pixel value are estimated.

FIG. 2 is a schematic block diagram of an image interpolation apparatus 100 using characteristics of a CFA pattern according to an embodiment of the present invention.

2, the image interpolating apparatus 100 includes an edge direction detecting unit 110, a weight calculating unit 120, an interpolating unit 130, and an adjusting unit 140. As shown in FIG.

In one embodiment of the present invention, horizontal and vertical DAIDs are calculated directly from the CFA image to determine the interpolation direction. Generally, the artifacts in the demosaiced image appearing in the high frequency region are primarily caused by the aliasing in the R / B channel because the decimated R / B channels have half the number of pixels compared to the decimated G channel Because. Preferably, a high correlation between the color channels R, G, B exists, i.e., inter-channel correlation exists in the high-frequency region of the color image.

The high frequency components of the R, G, and B channels are large in the edge and texture regions. It is assumed that the positions of the edges are the same in the R, G, and B channels. Here, in order to estimate the edge direction and the directional interpolation weight, an absolute channel difference is used, which is calculated directly from the CFA pattern. The components along the center column shown in Fig. 1A alternate between G and R. Fig. 1b and 1c. At each pixel, the interpolation direction is calculated using neighboring pixel values along the horizontal and vertical directions, and the adaptive direction weight is estimated using the spatial correlation between neighboring pixels along the detected interpolation direction.

The edge direction detection unit 110 calculates the absolute channel difference values in the left, right, up and down directions on the basis of the center pixel according to the pattern characteristics of the R, G, and B pixels arranged in the color filter array, do. (D _H , D _V ) for the detection of the edge direction, where the absolute channel differences (D _H ^L , D _H ^R , D _V ^U , D _V ^D ) are used. For example, referring to the CFA pattern of FIG. 1A, the DAIDs (D _H , D _V ) are defined along the horizontal and vertical directions as shown in Equations 4 and 5, respectively.

Wherein, R _pq and G _{p, q,} R, and the color intensity value at the position (p, q) in the G channel, and, D _H ^L, D _H ^R, D _V ^U, D _V ^D is the left and right, respectively, And the absolute channel difference in the up and down directions, respectively. DAIDs (D _H , D _V ) are computed in a 5 × 5 window using R _{i, j} at the center of the CFA.

Figure Ib is similar except that only _{Rp, q} and _{Bp, q} are replaced. Therefore, DAIDs (D _H , D _V ) are calculated by replacing R _{p, q} with B _{p, q} in the above equations (4) and (5) using the CFA pattern in FIG. FIG using 1c CFA pattern _{of, DAIDs (D H, D V} ) of the G _{p, q} an R _{p, q} in the equation 4 and 5 change as G _{p, q,} from the equation (4) B _p, switching to _q, and calculates the change in the mathematical expression (5), the G _{p, q} with R _{p, q.}

The weight calculation unit 120 calculates directional weights in the left, right, up and down directions according to pattern characteristics of the R, G, and B pixels.

The interpolator 130 receives the left, right, upper, and lower pixels from the R, G, and B pixels according to the edge direction detected by the edge direction detecting unit 110, that is, the horizontal difference D _H and the vertical difference D _V color difference values of the direction, for example, ΔG _H ^L, ΔG _H ^R, ΔG _V ^U, ΔG _V ^D and weighted directional weights calculated in the calculation unit 120, i.e., γ _H ^L, γ _H ^R, γ _V ^U, and interpolates unknown pixels using? _V ^D.

After detecting the edge direction, an adaptive interpolation weight is derived, and the unknown pixel value is adaptively interpolated using the direction interpolation weight and color difference values.

In FIG. 1A, there is an R pixel at the center of a 5x5 Bayer CFA pattern, and an unknown G pixel value, G ' _{i, j,} is interpolated as shown in Equation 6 below.

Here, ΔG _H ^L , ΔG _H ^R , ΔG _V ^U , and ΔG _V ^D are color difference values defined in the left, right, up, and down directions, respectively, and are calculated as shown in the following Equations 7 to 10, respectively. Referring together to Fig. 3, ΔG _H ^L is the left direction 310, as shown in the following and take advantage of Equation 7, ΔG _H ^R is using the following equation (8) as a right direction (320), ΔG _V shown in Figure 3 ^U uses the following equation 9 as the upper direction 330 shown in FIG. 3 and ΔG _V ^D is calculated using the following equation 10 as the lower direction 340 shown in FIG. 3, respectively.

In Equation (6), γ _H ^L , γ _H ^R , γ _V ^U , and γ _V ^D represent direction interpolation weights defined in left, right, up and down directions, respectively, do.

1B is similar to FIG. 1A except that R _{p, q} and B _{p, q} are changed. Thus, B pixels are 5 × 5 in the Figure 1b center of the Bayer CFA pattern, changing the G pixel values of the image, G _{'i, j} is R _{p, q} with B _{p, q,} using the equation (6) to 14 .

Interpolation section 130 performs R pixel and B pixel interpolation after G pixel interpolation. For example, in FIG. 1B, where the B pixel is the center of a 5x5 Bayer CFA pattern, the unknown R pixel value, _{R'i, j,} is interpolated using:

Here, ΔR _H ^L , ΔR _H ^R , ΔR _V ^U , and ΔR _V ^D are the color difference values defined in the left, right, top, and bottom directions, respectively, and similarly, after rotating the CFA pattern clockwise by 45 degrees do. Referring to FIG. 4A together, left, right, up and down directions are defined as shown and calculated as shown in the following equations (16) to (19), respectively.

In Equation (15), γ _H ^L , γ _H ^R , γ _V ^U , and γ _V ^D represent directionally interpolated weights similarly defined in the left, right, up, and down directions, respectively, .

In one embodiment of the present invention, R pixels are interpolated using the directional interpolation weights of FIG. 1B calculated according to Equations 20 through 23 above.

Fig. 1A is similar to Fig. 1B except that B _{p, q} and R _{p, q} are replaced. Thus, in FIG. 1A, in which the R pixel is the center of a 5 × 5 Bayer CFA pattern, the unknown B pixel value (B ' _{i, j} ) is obtained by replacing B _{p, q} with R _{p, q} , And is similarly estimated.

In FIG. 1C, for the R / B pixel interpolation, the direction weights? _H ^L ,? _H ^R ,? _V ^U ,? _V ^D are R _{p, q} is G _{p, q} , G _p, 12, B _{p, q} are substituted for R _{p, q} in Equations (13) and (14), and are estimated using Equations (11) to (14). The unknown R pixel value (R ' _{p, q} ) is interpolated using Equation (15) and the color difference value and the direction weight, the color difference value being defined as:

The image B pixel, B _{'p, q} is the equation (15), and color difference values of the R pixels ΔR _H ^L, ΔR _H ^R, ΔR _V ^U, ΔR _V ^D color difference values of the B pixels ΔB _H ^L, ΔB _H ^R, ΔB _V ^U, ΔB respectively replaced with _V ^D, the equation 24 to 27 of R _{'p, q} the equation (24), and from 25 to B _{p, q} equation 26 and 27, B' _{p, q} . As shown in Figs. 4C and 4D, the left, right, up and down directions are defined differently according to the center pixel.

The adjustment unit 140 adjusts the interpolated pixel value by adding the high frequency component value of the known pixel to the interpolated pixel value. The adjustment unit 140 may be optionally included in the image interpolating apparatus 100.

In one embodiment of the invention, the demapped image is enhanced by adjustment processing. The di-hatched image, e.g., R ', G', B ', may include eye-popping artifacts such as zipper artifacts, blurring, false color, Artifacts appear primarily in the high frequency domain. Accordingly, the adjustment unit 140 adjusts the demosaiced image using the high-frequency component having the inter-channel correlation of the R / G / B channel.

는 다음 수학식 28 및 29와 같다.For example, a G pixel is updated by adding a high frequency component of R or B pixels, which is a known pixel value in the CFA pattern. For example, in Figure 1A where the R pixel is the center of a 5x5 Bayer CFA pattern, the G / B components are adjusted. The adjusted G component,

Is expressed by the following equations (28) and (29).

를

로, G'_i,j를 B'_i,j로 대체하고, 동일하게 조정할 수 있다. 예를 들면 5×5 베이어 CFA 패턴의 센터가 각각 B와 G 픽셀들인 도 1b 및 1c에서, R/G 및 R/B 컴포넌트들이 유사한 방식으로 조정된다. 이러한 조정 단계를 수행한 후에, 조정부(140)는 디모자이싱된 이미지, 즉

,

,

를 출력한다. Here, G _{'i, j} denotes the interpolated G pixel in from the adaptive interpolation step ^{(i, j), R h} i, j is a high-frequency component of the R channel, R ^l _{i, j} is the detected horizontal or Means a low frequency component filtered by a 3-tap 1-D filter [1/3, 1/3, 1/3] along the vertical edge direction. Similarly, the B pixel can be expressed in Equations 28 and 29

To

, G ' _{i, j} can be replaced with B' _{i, j} , and the same adjustment can be made. For example, in Figures 1b and 1c, where the centers of the 5x5 Bayer CFA patterns are B and G pixels respectively, the R / G and R / B components are adjusted in a similar manner. After performing this adjustment step, the adjustment unit 140 adjusts the demosaiced image, i.e.,

,

,

.

Figures 5A through 5H are diagrams for comparing visualized quality of interpolated images according to one embodiment of the present invention and interpolated images according to other conventional demagnification algorithms.

 5A is an original image, FIGS. 5B through 5D are images demagnified according to a conventional demagnification algorithm, and FIG. 5E is a demosaiced image according to an embodiment of the present invention.

As can be seen intuitively from the figures, in the images of FIGS. 5B to 5D, the color edges that have been demapped by interpolation are blurred, or a zipper effect appears along the vertical color edge lines. The demapped image shown in Figure 5e shows no zip effect and satisfactory results.

6A to 6C are diagrams for comparing an interpolated image according to an embodiment of the present invention and an interpolated image according to other conventional demagnification algorithms in terms of a pixel intensity profile.

Figs. 6A to 6C show one-dimensional intensity profiles of R, G, B channels along 50 horizontal lines in the test image of Fig. 5A. A color edge is a sharp boundary between two distinct colors (R and B, G = 0). To determine the direction of interpolation, an embodiment of the present invention calculates horizontal and vertical DAIDs directly from the CFA image. Three color components, R / G / B, are interpolated along the horizontal and vertical directions detected in consideration of DAIDs (D _H , D _V ). The image interpolation method according to an embodiment of the present invention preserves sharp color edges in the resulting image without blurring and zip artifacts.

In addition, the image interpolation according to an exemplary embodiment of the present invention is larger than the conventional algorithms having different Peak Signal and Noise Ratio (PSNR) and Structural Similarity (SSIM), and has excellent visual quality. In particular, the PSNR of the proposed method for images with multiple color edges has a larger value compared to other conventional demosaicing methods. In addition, the SSIM for images with color sharp edges and a natural backgorund is larger than other conventional demagnetizing methods.

7 is a view for explaining an image interpolation method using characteristics of a CFA pattern according to another embodiment of the present invention.

Referring to FIG. 7, in step 700, a color filter array image is input. Here, the image is an image before being demagnetized, consisting of a certain arrangement of R, G, B. In step 702, the edge direction is detected using the DAID (D _H , D _V ) values in the horizontal and vertical directions. In step 704, weights are calculated. In step 706, the unknown pixel value is interpolated using the weight and color difference values calculated in step 704 according to the edge direction detected in step 702. [ In step 706, the interpolated pixel values are adjusted using the high-frequency components of each pixel.

In a preferred embodiment of the present invention, an image interpolation method and apparatus using the characteristics of the CFA pattern are provided. That is, edge direction and direction weights are calculated directly from the CFA image using DAIDs. Experimental results using a large number of different images reveal that it is an effective method of de-mosaicing the zipper effect and color artifacts in terms of PSNR, SSIM and visual quality.

Although a 5 × 5 Bayer CFA pattern has been described in the preferred embodiment of the present invention, it is not limited to a 5 × 5 pixel number, and image interpolation can be performed by using more pixels.

Meanwhile, the present invention can be embodied in computer readable code on a computer readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored.

Examples of the computer-readable recording medium include a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device and the like, and also a carrier wave (for example, transmission via the Internet) . In addition, the computer-readable recording medium may be distributed over network-connected computer systems so that computer readable codes can be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the present invention can be easily deduced by programmers skilled in the art to which the present invention belongs.

The present invention has been described above with reference to preferred embodiments. It will be understood by those skilled in the art that the present invention may be embodied in various other forms without departing from the spirit or essential characteristics thereof. Therefore, the above-described embodiments should be considered in an illustrative rather than a restrictive sense. The scope of the present invention is indicated by the appended claims rather than by the foregoing description, and all differences within the scope of equivalents thereof should be construed as being included in the present invention.

Figs. 1A to 1C show a 5x5 Bayer CFA pattern in which the R, B, and G pixels are center (i, j) pixels, respectively.

FIG. 2 is a schematic block diagram of an image interpolation apparatus 100 using characteristics of a CFA pattern according to an embodiment of the present invention.

Fig. 3 is a diagram for explaining a method of calculating a color difference value in a CFA pattern to interpolate an unknown pixel. Fig.

4A to 4D are diagrams for explaining a method of calculating a color difference value in a CFA pattern in order to interpolate an unknown pixel.

FIGS. 5A through 5E are diagrams for comparing visualized quality of interpolated images according to one embodiment of the present invention and interpolated images according to other conventional demagnification algorithms.

6A to 6C are diagrams for comparing an interpolated image according to an embodiment of the present invention and an interpolated image according to other conventional demagnification algorithms in terms of a pixel intensity profile.

7 is a view for explaining an image interpolation method using characteristics of a CFA pattern according to another embodiment of the present invention.

Description of the Related Art

100: image interpolating device 110: edge direction detecting section

120: Weight calculation unit 130:

140:

Claims (19)

Detecting edge directions by calculating absolute difference values in the left, right, up and down directions based on the center pixel according to the pattern characteristics of the R, G, and B pixels arranged in the color filter array; Calculating directivity weights in the left, right, up and down directions according to pattern characteristics of the R, G, and B pixels, respectively; And Interpolating unknown pixels using the color difference values in the left, right, up and down directions from the R, G and B pixels according to the detected edge direction and the calculated directional weights, Wherein the absolute channel difference value is a difference between a color intensity value of the first channel and a color intensity value of a neighboring pixel of the second color channel on one side of the center pixel, Is calculated by subtracting the absolute value of the difference between the color intensity value of the second color channel and the color intensity value of the neighboring pixel of the second color channel. The method according to claim 1, The absolute channel-to-channel difference values, And the difference values of neighboring pixel values in the left, right, upper, and lower directions of the color filter array pattern. 3. The method of claim 2, Wherein the detecting step comprises: (D _H ), which is the left-right difference value and the vertical difference (D _V ), which is the difference value in the up-down direction from the absolute-channel difference values in the left, right, And calculating the color filter array pattern using the characteristic of the color filter array pattern. The method of claim 3, Wherein the detecting step comprises: If the center pixel of the color filter array pattern is R _{i, j} , Wherein the horizontal difference (D _H ) and the vertical difference (D _V ) are calculated using the following equations (4) and (5). &Quot; (4) "

&Quot; (5) "

(Wherein, R _pq and G _{p, q} in the R and G pixels, respectively (p, q) the color intensity value is, D _H ^L, D _H ^R, D _V ^U, D _V ^D are respectively left and right in, Absolute difference between the upper and lower channels) 5. The method of claim 4, Wherein the interpolation step comprises: Wherein when the center pixel of the color filter array pattern is R _{i, j} , the unknown pixel G is interpolated using the following equation (6). &Quot; (6) "

(Wherein, G _{'i, j} is the interpolated G pixel value, ΔG _H ^L, ΔG _H ^R, ΔG _V ^U, ΔG _V ^D are respectively left and right, deulyigo phase, the downward color difference value, γ _H ^L , γ _H ^R , γ _V ^U , and γ _V ^D are directional weights in the left, right, up, and down directions, respectively) The method of claim 3, Wherein the interpolation step comprises: Interpolates G pixels among unknown pixels, and interpolates the remaining unknown pixel values based on the interpolated G pixel values (G '). The method according to claim 6, Wherein the interpolation step comprises: When the center pixel of the color filter array pattern is B _{i, j} , the unknown R pixel is interpolated according to the following expression (15). &Quot; (15) "

(Wherein, R _{'i, j} and G' _{i, j} is deulyigo pixel value interpolation, respectively, ΔR _H ^L, ΔR _H ^R, ΔR _V ^U, ΔR _V ^D are each left, right, up, the downward color difference Values,

,

,

,

Γ _H ^L , γ _H ^R , γ _V ^U , and γ _V ^D are the directional weights in the left, right, up, and down directions, respectively. 8. The method of claim 7, When the center pixel of the color filter array pattern is G _{i, j} , the color difference values in the left, right, up, and down directions in Equation (15)

,

,

,

Wherein the color filter array pattern is formed on the substrate. The method according to claim 1, Further comprising the step of adjusting the interpolated pixel value by adding the high frequency component value of the known pixel to the interpolated pixel value. 10. A recording medium on which a program for causing a computer to execute the method according to any one of claims 1 to 9 is recorded. An edge direction detecting unit for detecting edge directions by calculating difference values between absolute channels in the left, right, up and down directions with respect to the center pixel according to pattern characteristics of R, G, and B pixels arranged in the color filter array; A weight calculation unit for calculating directional weights in left, right, up and down directions according to pattern characteristics of the R, G, and B pixels; And And an interpolator interpolating unknown pixels using color difference values in the left, right, upper, and lower directions from the R, G, and B pixels according to the detected edge direction and the calculated directional weights, Wherein the absolute channel difference value is a difference between a color intensity value of the first channel and a color intensity value of a neighboring pixel of the second color channel on one side of the center pixel, And the absolute value of the difference between the color intensity value of the first color channel and the color intensity value of the neighboring pixel of the second color channel. 12. The method of claim 11, The absolute channel-to-channel difference values, And the difference values of neighboring pixel values in the left, right, upper, and lower directions. 13. The method of claim 12, Wherein the edge direction detecting unit comprises: The horizontal difference D _{H, which} is the difference between the left and right directional absolute values, and the vertical difference D _V , which is the difference value between the left and right directions, Wherein the color filter array pattern is formed on the substrate. 14. The method of claim 13, Wherein the edge direction detecting unit comprises: If the center pixel of the color filter array pattern is R _{i, j} , Wherein the horizontal difference (D _H ) and the vertical difference (D _V ) are calculated using the following equations (4) and (5). &Quot; (4) "

&Quot; (5) "

(Wherein, R _pq and G _{p, q} in the R and G pixels, respectively (p, q) the color intensity value is, D _H ^L, D _H ^R, D _V ^U, D _V ^D are respectively left and right in, Absolute difference between the upper and lower channels) 15. The method of claim 14, Wherein the interpolator comprises: And the center pixel of the color filter array pattern is R _{i, j} , the unknown pixel G is interpolated using the following equation (6). &Quot; (6) "

(Wherein, G _{'i, j} is the interpolated G pixel value, ΔG _H ^L, ΔG _H ^R, ΔG _V ^U, ΔG _V ^D are respectively left and right, deulyigo phase, the downward color difference value, γ _H ^L , γ _H ^R , γ _V ^U , and γ _V ^D are directional weights in the left, right, up, and down directions, respectively) 14. The method of claim 13, Wherein the interpolator comprises: Interpolates G pixels among unknown pixels, and interpolates the remaining unknown pixel values based on the interpolated G pixel values (G '). 17. The method of claim 16, Wherein the interpolator comprises: Wherein when the center pixel of the color filter array pattern is B _{i, j} , the unknown R pixel is interpolated according to the following expression (15). &Quot; (15) "

(Wherein, R _{'i, j} and G' _{i, j} is deulyigo pixel value interpolation, respectively, ΔR _H ^L, ΔR _H ^R, ΔR _V ^U, ΔR _V ^D are each left, right, up, the downward color difference Values,

,

,

,

Γ _H ^L , γ _H ^R , γ _V ^U , and γ _V ^D are the directional weights in the left, right, up, and down directions, respectively. 18. The method of claim 17, When the center pixel of the color filter array pattern is G _{i, j} , the color difference values in the left, right, up, and down directions in Equation (15)

,

,

,

Wherein the color filter array pattern is formed on the substrate. 12. The method of claim 11, And adjusting the interpolated pixel value by adding the high-frequency component value of the known pixel to the interpolated pixel value.

Images