JP2005333418A - Color solid state imaging apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the problem that a color false signal is generated on an edge part between colors of high saturation in a conventional color solid state imaging apparatus. <P>SOLUTION: Respective G signals of all pixels in a pixel signal having the sensitivity of G light are generated by using a highly accurate interpolation processing circuit 14, color difference signals on the pixel positions of R and B light components are calculated, respectively, and after performing interpolation processing by interpolation processing circuits 18, 19 on the basis of the correlation of the G signals, color difference signals of all pixels are synthesized. Since the G signal and the edge position of two color difference signal are aligned each other, a color image of high image quality in which a color false signal is not included can be obtained even on the edge of colors of high saturation. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

The present invention relates to a color solid-state image pickup device using a solid-state image pickup device having a mosaic color filter composed of three colors used for a video movie, an electronic still camera, an electronic endoscope, and the like. Regarding the method.

Japanese Unexamined Patent Application Publication No. 2002-112276 is a technique for reducing color false signals of a conventional color solid-state imaging device. As shown in FIG. 12, G (green) pixels of the output video signal of the solid-state imaging device are subsampled, and interpolation processing is performed with high accuracy to obtain G signals for all the pixels. At the positions of R (red) and B (blue) pixels, interpolated G signals are subtracted to obtain color difference signals, and the obtained color difference signals are interpolated to generate color difference signals for all pixels. Then, matrix calculation is performed to obtain R, G, and B primary color signals or Y (luminance), Cb (blue difference), and Cr (red difference) three signals for all pixels. For this reason, since the color difference signal is 0 for an achromatic image, a color false signal is not generated even at an edge portion (a portion where the light and darkness changes suddenly), and a natural image is not required to lighten the color at the edge portion. It is to be obtained.

However, a color false signal is generated as follows at the edge portion of high saturation as shown in FIG. FIG. 13A shows a part of an image projected on the image sensor, and shows edges of green and magenta (red and blue additive color mixture). In FIG. 13B, R, G, and B pixels of the image sensor are shown as circles, squares, and pentagons, respectively, and signal values are shown as filled brightness. That is, in the green region, the white square G pixel has a signal value of 1, and the black circle, pentagonal R pixel, and B pixel have a signal value of 0. In the magenta area, the white circle and pentagonal R pixel and B pixel have a signal amount of 1, and the black square G pixel has a signal amount of 0. The boundary portion has an intermediate signal level and the signal amount is indicated by hatching.

The G pixel is subsampled, and interpolation processing is performed to obtain G signals of all the pixels in FIG. The G signal of all pixels is subtracted from the image sensor output video signal, and sub-sampled at the R pixel position and the B pixel position, respectively, and the RG image of FIG. 13D and the BG image of FIG. Generated. Hatching indicates the signal value of the color difference signal, with black corresponding to -1 and white corresponding to +1. The solid lines in FIGS. 13C to 13E indicate the edge shapes of the G signal and the two color difference signal images, respectively. Since the edge shape of the two color difference signals with coarse sampling is different from the edge shape of the original image, there is a problem that a false color signal is generated at the edge portion of high saturation as shown in FIG.
JP2002-112276

The above-described conventional color solid-state imaging device has a problem in that a color false signal is generated at an edge portion between colors with high saturation.

In order to solve the above problems, in the color solid-state imaging device of the present invention, a mosaic color filter in which a pixel having a first spectral sensitivity and a pixel having a second or third spectral sensitivity are arranged adjacent to each other. And generating an image signal of the first spectral sensitivity for all pixels of the solid-state image sensor by interpolation processing from an output signal of the solid-state image sensor, and correlating with the interpolated signal A correlation signal representing a strong direction is generated, a difference signal from the pixel signal of the first spectral sensitivity is generated at each pixel position of the second and third spectral sensitivities, and the second and third color lights are generated. A color difference pixel signal for all pixels is obtained from the corresponding difference signal based on the correlation signal.

By the means described above, the direction in which the pixel correlation of the interpolated image signal with the first spectral sensitivity is strong and the direction in which the pixel correlation of the color difference image signal with respect to the interpolated second and third spectral sensitivities are aligned are aligned. As a result, the edge of the image signal with the first spectral sensitivity matches the edge of the color difference image signal with respect to the second and third spectral sensitivities. Therefore, it is possible to reduce the color false signal generated at the edge portion between the highly saturated colors.

A solid-state imaging device in which a pixel having the first spectral sensitivity and a pixel having the second or third spectral sensitivity are arranged adjacent to each other is used. First, an image signal having the first spectral sensitivity for all pixels is generated, and a correlation signal representing a direction having a strong correlation with the interpolated signal is generated. A difference signal from the generated pixel signal of the first spectral sensitivity is generated at each pixel position of the second and third spectral sensitivity. When generating chrominance pixel signals for all the pixels corresponding to the second and third color lights by interpolation processing, by selecting a pixel for calculating an interpolation value according to the correlation signal, the first spectral sensitivity signal and The image edges of the two color difference signals can be matched. Accordingly, it is possible to realize a color solid-state imaging device in which almost no color false signal is generated at the edge portion between colors with high saturation.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of a color solid-state imaging device according to the first embodiment of the present invention. Reference numeral 1 denotes a solid-state image sensor, which is driven by an image sensor drive pulse generated by a timing generator 10 via a drive circuit 11 and outputs a video signal at a predetermined timing. As shown in the figure, this solid-state imaging device 1 is provided with a mosaic color filter 2 in which G pixels are arranged in a checkered pattern, and R and B fill G in every row. And BGBG... Are output line by line. Although not shown in the drawing, this signal can be used to refer to the signal amount of the pixel to be processed and its peripheral pixels by using a plurality of delay lines and storage means such as a frame memory. Hereinafter, description will be made assuming that the interpolation position is an R pixel as shown in FIG.

The interpolation processing means 3 outputs the pixel signal as it is for the G pixel position of the solid-state imaging device 1, and for the position of the R pixel or B pixel, according to the flowchart of FIG. The value and correlation direction are determined and output. After initializing the correlation direction, processing is performed based on the light / dark order pattern of four G pixels (GU, GR, GL, GD) adjacent to the interpolation position. A set of four circles arranged in a diamond shape in FIG. 3 indicates four pixels adjacent to the interpolation position, a white circle indicates the brightest pixel, and a black circle indicates the darkest pixel. The hatched circles indicate intermediate brightness, and the brightness varies depending on the darkness of the hatching.

The light / dark order pattern P1 corresponds to the right-downward boundary line when the signal values of GU and GR are equal and the signal values of GL and GD are equal. The direction in which the correlation is strong cannot be determined from the adjacent G pixel, but if B1 to B4 have a right-downward boundary line, it can be determined by the strong correlation of the B pixel. FIG. 4 shows the correlation between the position of the lower right boundary line and the signal values B1 to B4. FIG. 4A shows a case where the boundary line is at the lower left of the R pixel, and the correlation between the signal values of B1, B2, and B4 is strong, and the correlation is strong in the upper right direction. FIG. 4B shows a case where the boundary line is at the upper right of the R pixel, and the correlation between the signal values of B1, B3, and B4 is strong, and the correlation is strong in the lower left direction. FIG. 4C shows a case where the boundary line is in the R pixel, and there is no direction in which the correlation is strong. Therefore, when the direction in which the signal values of B1 to B4 are strongly correlated is the upper right or the lower left, the interpolation value of the R pixel position and the correlation direction information are determined in the direction of the strong B correlation. In other cases, processing is performed when the next correlation has no directionality.

When the correlation has no directionality, the average of the signal values of the four adjacent G pixels or the average of the intermediate binary values is used as the interpolation value. When the difference between the signal amount of the R pixel and the interpolation value is small and the color difference signal at the B1 and B2 pixel positions is also small, it is considered to be a light-colored part, so an oblique fringe component of the signal values of B1 to B4 is added, Make a broadband signal.

The light / dark order pattern P2 is processed on the assumption that the left and right sides of the light / dark order pattern P1 are interchanged.

The light / dark order pattern P3 is a case where the four adjacent G pixels have substantially the same signal value, and the correlation is not directional, and the above-described correlation is not directional.

The light / dark order pattern P4 is a case where the signal values on the upper and lower sides and the left and right sides of the four adjacent G pixels are close to each other, and there is a vertical line or a horizontal line at the R pixel position. Of the four G pixel signal values, only one greatly different pixel can be determined as a horizontal stripe when it is above or below, and as a vertical stripe when it is left or right. In other cases, the determination is made based on the stripe components of the diagonally adjacent pixels B1 to B4. When the vertical stripe is determined, the interpolation value is the average of the upper and lower pixels (GU, GD), and the correlation direction is determined depending on which of the signal values of the left and right G pixels (GL, GR) is closer. If it is determined as horizontal stripes, the interpolation value is the average of the left and right G pixels (GL, GR), and the correlation direction is determined depending on which of the interpolation values is closer to the signal values of the upper and lower G pixels (GU, GD). To do. If neither vertical stripes nor horizontal stripes are determined, processing is performed when the above-described correlation has no directionality.

The light / dark order pattern P5 is a case where the brightness of the left and upper G pixels (GL, GU) is the brightest pixel and the darkest pixel, and the brightness changes sequentially in the direction of rotation around the R pixel. As an image having such a pattern, there may be a case where a shadow due to a step is generated in the upper left direction. Therefore, the direction in which the correlation between the R pixel positions is strong is the lower right. However, since there is no clear edge at the R pixel position, the interpolation value is determined by the above-described processing when the correlation has no direction.

Since the light and dark order patterns P6 to P8 are obtained by rotating the light and dark order pattern P5, the correlation direction is set to the direction of the two pixels having intermediate brightness, and the interpolation value is based on the above processing when the correlation has no directionality. Decide.

The light / dark order pattern P9 is a case where the signal values of the upper and lower G pixels are the brightest and darkest pixels, and the signal values of the left and right G pixels are intermediate values. An image having such a pattern is an edge image in which a boundary line extends in the horizontal direction. The interpolation value is an average of the signal values of the left and right pixels (GL, GR), and the interpolation value is the upper or lower G pixel. The correlation direction is determined based on the proximity of the signal value.

The light / dark order pattern P10 is a case where the signal values of the upper and lower G pixels are the brightest and darkest pixels, and at least one of the signal values of the left and right G pixels is the same as the signal value of the lower G pixel. This is an edge image in which the boundary line extends in the horizontal direction, and the interpolation value is the average of the signal values of the left and right pixels (GL, GR), and the interpolation value is always close to the signal value of the lower G pixel (GD). , Set the correlation direction down.

The light / dark order pattern P11 is a case where the signal values of the upper and lower G pixels are the brightest and darkest pixels, and at least one of the signal values of the left and right G pixels is the same as the signal value of the upper G pixel. This is an edge image whose boundary line extends in the horizontal direction, and the interpolation value is the average of the signal values of the left and right pixels (GL, GR), and the interpolation value is always close to the signal value of the upper G pixel (GU). , Set the correlation direction up.

The light / dark order pattern P12 is a case where the signal values of the left and right G pixels are the brightest and darkest pixels, and the signal values of the upper and lower G pixels are intermediate values. An image having such a pattern is an edge image in which a boundary line extends in the vertical direction. The interpolation value is an average of the signal values of the upper and lower pixels (GU, GD), and the interpolation value is the left or right G pixel. The correlation direction is determined based on the proximity of the signal value.

The light / dark order pattern P13 is a case where the signal values of the left and right G pixels are the brightest and darkest pixels, and at least one of the signal values of the upper and lower G pixels is the same as the signal value of the left G pixel. This is an edge image in which the boundary line extends in the horizontal direction, the interpolation value is the average of the signal values of the upper and lower pixels (GU, GD), and the interpolation value is always close to the signal value of the left G pixel. Set to the left.

The light / dark order pattern P14 is a case where the signal values of the left and right G pixels are the brightest and darkest pixels, and at least one of the signal values of the upper and lower G pixels is the same as the signal value of the right G pixel. This is an edge image in which the boundary line extends in the horizontal direction, the interpolation value is the average of the signal values of the upper and lower pixels (GU, GD), and the interpolation value is always close to the signal value of the right G pixel. Set to the right.

As described above, the interpolation value and correlation direction of the G signal at the R pixel position can be obtained with high accuracy by analyzing the light / dark order pattern of the G pixel adjacent to the R pixel and the signal values of the G pixel and B pixel surrounding the R pixel. Can be determined.

Further, the interpolation value and correlation direction of the G signal at the B pixel position can be determined with high accuracy by replacing the R pixel and the B pixel in the R pixel position determination method, and G with very little blur for all the pixels. A signal is obtained.

Therefore, the interpolation processing means 3 outputs a G signal approximated to the RB pixel position with high accuracy. Although not shown, a solid-state imaging device output at the same position as the G signal output from the interpolation processing means 3 is supplied to the subtractor 4 by a storage means such as a delay means or a frame memory, and the G signal is subtracted and thinned out. When the signal at the R pixel position is sampled by the means 5, an RG signal is obtained, and when the signal at the B pixel position is sampled by the thinning means 6, a BG signal is obtained.

The interpolation processing means 7 generates an RG signal at the B pixel position according to the correlation direction signal, and then generates an RG signal at the G pixel position. FIG. 5 shows the relationship between the correlation direction signal at the B pixel position and the RG signal to be generated. The white arrow indicates the correlation direction output from the interpolation processing means 3, and the arrow line with a numerical value indicates the signal synthesis ratio. For example, FIG. 5A shows a case where the correlation is strong in the upper left direction, where 0.5 times the RG signal of the pixel adjacent to the upper left and 0 for each of the RG signals of the pixels adjacent to the lower left and upper right. By adding .25 times, an RG signal having a strong correlation with the upper left direction is generated. 5 (c), 5 (g), and 5 (i) are cases where the directions of strong correlation are the upper right, lower left, and lower right, respectively, and the RG signals of 3 pixels are added at the same ratio. To generate an interpolation value. FIG. 5B shows a case where the correlation is strong in the upward direction, and 0.5 times each of the RG signals of the pixels adjacent to the upper left and upper right are added to generate an RG signal having a strong correlation with the upward direction. . FIG. 5D, FIG. 5F, and FIG. 5H show the cases where the strong correlation directions are left, right, and bottom, respectively, and an interpolation value is generated by averaging two-pixel RG signals. FIG. 5E shows a case where a direction with strong correlation cannot be specified, and an RG signal having no correlation direction is generated by taking an average of RG signals of four pixels diagonally adjacent to each other. Further, the RG signal at the G pixel position is calculated by, for example, averaging the four pixels of the RG signals at the adjacent R pixel position and the B pixel position or the average of the binary values.

The interpolation processing unit 8 generates the BG signal at the R pixel position in accordance with the correlation direction signal in the same manner as the interpolation processing unit 7, and then generates the BG signal at the G pixel position, and the BG signal at all the pixel positions. Get.

FIG. 6 shows an image generation process of an edge portion between high-saturation colors of green (signals only for G pixels) and magenta (signals are present for R pixels and B pixels) as a specific example. FIG. 6A shows the correspondence between the edge between the high-saturation color and the pixel, and green light is incident on the left and magenta light is incident on the right. The R, G, and B pixels of the image sensor are indicated by circles, squares, and pentagons, respectively. FIG. 6B shows the signal amount of each pixel by the brightness of the fill. That is, in the green region, the white square G pixel has a signal value of 1, and the black circle, pentagonal R pixel and B pixel have a signal value of 0. In the magenta area, the white circle and pentagonal R pixel and B pixel have a signal amount of 1, and the black square G pixel has a signal amount of 0. The boundary portion has an intermediate signal level and the signal amount is indicated by hatching.

FIG. 6C shows the amount of signal at the G pixel position indicated by a square and the direction of strong correlation at the R / B pixel positions. A blank R / B pixel position indicates that there is no strong correlation direction. An arrow indicates a direction having a strong correlation, and there is a strong correlation in the outer direction on both sides along the color boundary line. At the pixel position with strong left or right correlation, the interpolation value is determined by averaging the signal values of the G pixels adjacent vertically, and at the pixel position where the strong correlation direction cannot be specified, the signal value of the G pixel adjacent vertically, horizontally is determined. The interpolation value is determined on average, and the G signal of all the pixels in FIG. 6D reproduces the green light pattern of the incident light image with high accuracy.

The subtracting means 4 subtracts the G signal of all pixels from the image sensor output video signal, and generates an RG signal and a BG signal at the R pixel position and the B pixel position, respectively. The thinning means 5 and 6 extract the signals at the R pixel position and the B pixel position, respectively, to generate the RG image in FIG. 6E and the BG image in FIG. Hatching indicates the signal value of the color difference signal, with black corresponding to -1 and white corresponding to +1. The interpolation processing means 7 first determines an interpolated value of the RG signal based on the correlation information of the G signal at the B pixel position. FIG. 6 (f) shows the RG signal value at this point. In the B pixel position having a strong correlation with the left or right, the combination with the averaged RG signal is enclosed for easy understanding. Show. The center line of the edge of the RG signal is a straight line in the figure, and is the same as the center line of the edge of the reproduced G signal. The RG signal at the G pixel position is generated from the RG signals at the adjacent R pixel position and the B pixel position, and the center lines of the edges of the G signal image and the RG signal image of all the pixels coincide. Similarly, FIG. 6H shows a stage where the interpolated value of the BG signal at the R pixel position is determined. The center line of the edge of the BG signal is a straight line in the figure, and is the same as the center line of the reproduced G signal and the edge.

Since the edge positions of the two color difference signals coincide with the G signal of all the pixels generated in this way, even if the color signal is converted into the three color signals required by the matrix means 9, the edge between the high-saturation colors Therefore, it is possible to realize a color solid-state imaging device capable of obtaining a high-quality image that does not generate a false color signal.

FIG. 7 is a block diagram of a color solid-state imaging device according to the second embodiment of the present invention. Blocks having the same functions as those in FIG. The difference is that the pixels of the solid-state imaging device 1 are not a matrix arrangement but a checkered arrangement, and the three complementary color filters (C (cyan) and Y) that the mosaic color filter blocks the three primary colors of light respectively. (Yellow) · M (magenta)) and correlation detection means 22 and 23 are provided.

The output of the solid-state imaging device 1 having the checkered array of pixels can be referred to the signal amount of the processing target pixel and its peripheral pixels by using a plurality of storage means such as delay lines and frame memories. The interpolation processing means 3 outputs the pixel signal as it is for the Y pixel position of the solid-state imaging device 1, and for the position of the C pixel or M pixel, the interpolation value and the correlation direction with high accuracy are shown according to the flowchart of FIG. Determine and output. Since this interpolation process is the same as the interpolation process of FIG. 3 except that the adjacent pixel position is rotated by 45 degrees, a description thereof will be omitted.

The Y signal output from the interpolation processing unit 3 and the output of the solid-state image sensor at the same position are supplied to the subtractor 4 by a storage unit such as a delay unit or a frame memory (not shown), and a difference signal is output. When the signal at the C pixel position is sampled by the thinning means 5, a CY signal, that is, a BR signal is obtained. When the signal at the M pixel position is sampled by the thinning means 6, an MY signal, that is, a BG signal is obtained.

The correlation detection means 12 is supplied with a BG signal at a pixel position indicated by a circle in FIG. 9 generated by the thinning means 5 by a storage means such as a delay means or a frame memory not shown. The correlation detection means 12 is closer to the BG signal value at the center pixel position CO, which is closer to the BG signal value at the pixel position CU or CH, or closer to the BG signal value at the pixel position CI or CT. First, determine. For example, if the pixel position CU and CT are close to each other, it is determined that the correlation is strong in the right direction. If it is not close to the pixel position CU but close to either CI or CT, the correlation is strong in the upper right direction. If it cannot be determined that the pixel position is close to any of the pixel positions CU, CH, CU, and CT, it is determined that there is no strong correlation.

The correlation processing unit 7 receives outputs from the interpolation processing unit 3 and the correlation detection unit 12 as correlation direction signals. The correlation direction information generated by the interpolation processing unit 13 determined based on the neighboring pixel information. Priority. For this reason, when the correlation detection means 12 is implemented as software, the correlation detection means 12 may be implemented only when the interpolation processing means 3 cannot identify a strong correlation direction. The interpolation processing means 7 determines the interpolation value by weighting and adding the BR color difference signals at the M pixel positions adjacent in the vertical and horizontal directions at the signal ratio shown in FIG. 10 based on the correlation information. White arrows indicate correlation direction information, and arrow lines with numerical values indicate signal synthesis ratios. For example, FIG. 10A shows a case where the correlation is strong in the upper left direction. By adding 0.5 times each of the BR signals of the pixels adjacent to the left and the upper side, BR having a strong correlation with the upper left direction is shown. Generate a signal. FIG. 10C, FIG. 10G, and FIG. 10I are the cases where the directions of strong correlation are the upper right, lower left, and lower right, respectively, and an interpolation value is generated by averaging two-pixel BR signals. To do. FIG. 10B shows a case where the correlation is strong in the upward direction, 0.5 times the BR signal of the pixel adjacent to the upper side, and 0.25 each of the BR signal of the pixel adjacent to the left and right. Double the signal to generate a BR signal having a strong correlation with the upward direction. FIGS. 10 (d), 10 (f), and 10 (h) show the cases where the directions of strong correlation are left, right, and bottom, respectively, and three-pixel BR signals are added at the same ratio and interpolated. Generate a value. FIG. 10E shows a case where a direction having a strong correlation cannot be specified. An average of the BR signals of four pixels adjacent vertically and horizontally is taken to generate a BR signal having no correlation direction. In addition, the BR signal at the G pixel position is calculated by, for example, averaging the four pixels of the BR signals at the adjacent C pixel position and the M pixel position, or averaging the intermediate binary values.

Correlation detection means 13 and interpolation processing means 8 generate a BG signal at the C pixel position according to the correlation direction, and then generate a BG signal at the G pixel position in the same manner as the correlation detection means 12 and interpolation processing means 7. And BG signals at all pixel positions are obtained.

FIG. 11 shows an image generation process of an edge portion between high-saturation colors of green (signals are present in Y and C pixels) and red (signals are present in Y and M pixels) as a specific example. FIG. 11A shows the correspondence between edges and pixels between highly saturated colors. Red light is incident on the left and green light is incident on the right. The C, Y, and M pixels of the image sensor are indicated by pentagons, squares, and circles, respectively. FIG. 11B shows the signal amount of each pixel by the brightness of the fill. That is, in the red region, the white circle and square M and Y pixels have a signal amount of 1, and the black pentagonal G pixel has a signal amount of 0. In the green region, the white square and pentagonal Y pixel and C pixel have a signal value of 1, and the black circle M pixel has a signal value of 0. The boundary portion has an intermediate signal level and the signal amount is indicated by hatching.

FIG. 11C shows the signal amount at the Y pixel position indicated by a square. Since the Y pixel signal amounts are the same, the correlation direction at the C / M pixel position cannot be specified from the Y pixel signal, and is blank. Since the direction with strong correlation cannot be specified, the interpolation value is determined by the average of the signal values of Y pixels adjacent vertically and horizontally, and the Y signals of all the pixels in FIG. 11D are in a flat state.

The subtracting means 4 subtracts the Y signals of all the pixels from the image sensor output video signal, and the thinning means 5 and 6 extract the signals of the C pixel position and the M pixel position, respectively, and the BR of FIG. An image and a BG image shown in FIG. 11F are generated. Hatching indicates the signal value of the color difference signal, with black corresponding to -1 and white corresponding to 0. A white arrow indicates a direction of strong correlation detected by the correlation detection means 22 and 23. When interpolation processing is performed for each individual color difference signal, the dotted line becomes the center line of the edge, indicating that a color false signal is generated.

Since the interpolation processing unit 7 cannot specify the correlation direction information of the Y signal, the interpolation processing unit 7 determines the interpolation value of the BR signal at the M pixel position based on the correlation direction information of the BG signal detected by the correlation detection unit 12. FIG. 11G shows the BR signal value at this time. At the M pixel position having a direction with strong correlation, the combination of BR signals to be weighted and added is enclosed for easy understanding. The center line of the edge of the BR signal is indicated by a dotted line in the figure. Since the BR signal at the Y pixel position is generated from the BR signals at the adjacent C pixel position and the M pixel position, the center line of the edge of the BR signal of all the pixels is the dotted line in FIG. Is almost the same. Similarly, FIG. 11 (h) shows a stage where the interpolated value of the BG signal at the C pixel position is determined, and the center line of the edge of the BG signal is a dotted line in the figure, and the BR signal And almost the same as the edge centerline.

Thus, even for an image in which the signal value of the Y pixel does not change, the edge positions of the two color difference signals of all the generated pixels match. Further, when there is a change in the signal value of the Y pixel, the edge positions of the two color difference signals coincide with the Y pixels of all the pixels generated in the same manner as described in the first embodiment. Even when color conversion is performed by the matrix means 9 and output, since the edge shape is the same among the three signals, no false color signal is generated at the edge portion between the high-saturation colors, and a high-quality image is obtained. A color solid-state imaging device can be realized.

As described above, in the color solid-state imaging device of the present invention, one color signal obtained from the output of a solid-state imaging device having a mosaic color filter composed of three colors and two color difference signals are correlated with surrounding pixels. By performing the color difference signal interpolation processing so that the strong directions are the same, the change points of the three signals of the image edge can be matched, and the false color signal can be reduced.

The implementation form of the processing algorithm of the present invention is not limited. For example, hardware implementation is desirable for video movies and electronic endoscopes where real-time performance is important, and software implementation may be desired for electronic still cameras that require flexible image quality adjustment. Further, the three primary color filters and the three complementary color filters are exemplified as the color filters. For example, a combination of R (red), G (green), and C (cyan) may be used, and any desired spectral characteristics can be realized. Three colors can be selected, and a combination of ultraviolet, visible light, and infrared may be used, and there is no limited relationship with the pixel arrangement of the solid-state imaging device. In addition, the interpolation process for matching the direction in which the correlation between the color difference signals is strong is exemplified for the solid-state image sensor having the checkered pixel arrangement, but the relationship between the presence of the correlation matching process between the arrangement pixel of the solid-state image sensor and the color difference signal It is not limited to.

Block diagram of a color solid-state imaging device (Example 1) FIG. 10 is a diagram for explaining the positional relationship between peripheral images (Example 1) Flow chart of high-precision pixel interpolation processing (Example 1) FIG. 10 is a diagram for explaining the correlation between the position of the lower right boundary line and the signal value (Example 1). FIG. 10 is a diagram for explaining an interpolation value determination method for color difference interpolation processing (Example 1). FIG. 10 is a diagram for explaining the effect of color difference signal interpolation processing (Example 1); Block diagram of a color solid-state imaging device (Example 2) Flow chart of high-precision pixel interpolation processing (Example 2) Relationship diagram of pixel positions for obtaining correlation direction of color difference signals (Example 2) FIG. 10 is a diagram for explaining an interpolation value determination method for color difference interpolation processing (second embodiment). FIG. 10 is a diagram for explaining the effect of correlation direction matching processing between color difference signals (Example 2). Block diagram of a conventional color solid-state imaging device (conventional example) Diagram for explaining a color false signal generated at a high saturation edge (conventional example)

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Solid-state image sensor 2 Mosaic color filter 3 Interpolation processing means 4 Subtraction means 5 Thinning-out means 6 Thinning-out means 7 Interpolation processing means 8 Interpolation processing means 9 Matrix 10 Timing generator 11 Drive circuit 12 Correlation detection means 13 Correlation detection means 14 Thinning-out means

Claims (3)

A solid-state image sensor in which a pixel having the first spectral sensitivity characteristic and a pixel having either the second or third spectral sensitivity characteristic are arranged adjacent to each other, and an output signal of the solid-state image sensor To generate a signal of the first spectral sensitivity characteristic at the image position of the second or third spectral sensitivity with respect to all pixels of the solid-state imaging device, and in which direction the generated signal has the first spectral sensitivity. Generating a correlation direction signal indicating whether the pixel signal of the sensitivity characteristic is strongly correlated, and generating a difference signal from the pixel signal of the first spectral sensitivity at each of the pixel positions of the second and third spectral sensitivity, The two difference signals for all the pixels are generated by interpolation processing, and the first spectral sensitivity characteristic signal of all the generated pixels and the difference signals for the second and third spectral sensitivities are converted to a color image. Color to get signal In the solid-state imaging device, the difference signal corresponding to the third and second spectral sensitivities at the pixel positions of the second and third spectral sensitivities is generated based on at least the correlation direction signal. Solid-state imaging device. When the correlation direction signal at the pixel position of the second or third spectral sensitivity indicates that the correlation direction could not be specified, the generated second or third spectral sensitivity characteristic is indicated respectively. 2. The color solid-state imaging device according to claim 1, wherein an interpolated value of the difference signal corresponding to the third or second spectral sensitivity is generated based on a correlation between the corresponding difference signal and surrounding pixels. . A signal indicating a direction having a strong correlation with the pixel interpolation signal of the first spectral sensitivity characteristic signal is an 8-pixel surrounding the interpolation target pixel with priority given to the light / dark order pattern of the pixel signal of the four pixels adjacent to the interpolation target pixel. The color solid-state imaging device according to claim 1, wherein the color solid-state imaging device is determined based on a signal.

Images