KR20050103775A - Method for color interpolation of image sensor - Google Patents

Abstract

The present invention is to provide a color interpolation method of the image sensor that can reduce the false color, the present invention, to achieve the above object, the present invention, three of the eight pixels around any pixel adjacent to each other Setting eight boundary masks consisting of eight pixels; Setting an interpolation region; Determining a boundary by comparing a value 'B' obtained by multiplying an intensity value of a center pixel of the interpolation area by a preset mask value, and a value 'A' obtained by multiplying a corresponding value of each of the eight boundary masks and an intensity value of the pixel. ; Determining the center pixel as a bad pixel and performing bad pixel correction interpolation as the center pixel corresponds to a boundary in all eight directions; Performing false color prevention interpolation as the center pixel corresponds to a boundary in one to seven directions; And performing interpolation using an average as the center pixel does not correspond to a boundary in all directions.

Description

Color interpolation method of image sensor {METHOD FOR COLOR INTERPOLATION OF IMAGE SENSOR}

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image sensor, and more particularly, to a complementary metal oxide semiconductor (CMOS) image sensor, and more particularly, to a method of color interpolation for reducing false colors in a CMOS image sensor.

An image sensor is a device for taking an image by using a property in which a semiconductor reacts to light. That is, the device detects the brightness and wavelength of different light emitted from each subject and reads it as an electric value. It is the role of the image sensor to make this electrical value a signal-processing level.

1 is a block diagram schematically illustrating an image sensor.

Referring to FIG. 1, an image sensor includes a pixel array unit 10, an analog / digital converting unit 11, and a pixel line memory unit 12 including N × M (N, M is a natural number) unit pixels. And an image signal processor 13 and a control and external system interface 14.

The control and external system interface unit 14 controls the overall operation of the image sensor using a finite state machine (FSM), and serves as an interface to the external system. In addition, it has a batch register, which can be programmed for various internal operations, and controls the operation of the entire chip according to this program information.

The pixel array unit 10 has unit pixels arranged in horizontally N and vertically M pieces to maximize the light responsiveness. The pixel array unit 10 detects information on an image from the outside and is an essential part of the entire sensor. .

The analog / digital converting unit 11 converts the analog voltage sensed by each pixel of the pixel array unit 10 into a digital voltage so that the digital system can process it.

The pixel line memory unit 12 stores a digital voltage of a pixel converted by the analog / digital converter 11 and is composed of several lines to perform various functions of the image signal processor of the rear stage.

The image signal processor 13 may perform color interpolation, color correction, gamma correction, auto white balance, auto exposure, etc. according to the function of the image sensor. Based on the output values of the pixels stored in the pixel line memory unit 12, various functions for improving the performance of the image sensor are included.

Currently, the image sensor is mainly composed of VGA (640 × 480 pixels) and MEGA (1152 × 864 pixels), and in the future, it is mega class having more than 3 million pixels according to the generalization of digital still camera and camera mobile phone. Image sensors are expected to become commonplace. Image sensors that are used for various purposes are the most important factor in determining the price along with their size.

The image quality of the sensor depends on the characteristics of the pixel itself and the function of the image signal processing unit 13, and the interpolation of the image signal processing unit 13 is a large value that determines the image quality in an image sensor in which one pixel has one color intensity value. Element. Interpolation inevitably has a false color in the fine detail, which is a factor that greatly degrades the image quality of the sensor.

The pixel of the image sensor has an intensity value of one of red, green, and blue colors, and has a structure of reproducing the original color of the pixel through color interpolation. Color interpolation reproduces the color information of a pixel with information of one color by using the information of surrounding pixels with information of another color. In an image with little change, it is a problem to reproduce the color by using the information of surrounding pixels. However, there is a problem in that a false color is reproduced at the boundary line of the image where the color information of the pixel is greatly changed.

The present invention has been proposed to solve the above problems of the prior art, and an object thereof is to provide a color interpolation method of an image sensor that can reduce false colors.

In order to achieve the above object, the present invention comprises the steps of: setting eight boundary masks consisting of three pixels adjacent to each other among the eight pixels around any pixel; Setting an interpolation region; Determining a boundary by comparing a value 'B' obtained by multiplying an intensity value of a center pixel of the interpolation area by a preset mask value, and a value 'A' obtained by multiplying a corresponding value of each of the eight boundary masks and an intensity value of the pixel. ; Determining the center pixel as a bad pixel and performing bad pixel correction interpolation as the center pixel corresponds to a boundary in all eight directions; Performing false color prevention interpolation as the center pixel corresponds to a boundary in one to seven directions; And performing interpolation using an average as the center pixel does not correspond to a boundary in all directions.

According to the present invention, in performing color interpolation using a memory, an array consisting of intensity values of a center pixel and intensity values of pixels having the same color of surroundings is formed, and an array includes a mask for detecting edges in each direction. By applying, if the edge value in all directions is larger than the threshold value, the center pixel is judged as a bad pixel to perform bad pixel compensation interpolation, and one or more threshold values are set as the threshold value. If larger, false color compensation interpolation is performed. Through this method, it is possible to provide an image sensor which minimizes false colors occurring at the boundary of an image and improves image quality.

DETAILED DESCRIPTION Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the technical idea of the present invention. do.

An image sensor uses a color filter array for displaying one color to one pixel using a color filter for a pixel, and a Bayer color filter array is generally known. The arrangement of the Bayer color filter array is as follows.

G R G R G R ...

B G B G B G ......

The subsequent rows have the same repetitive structure.

FIG. 2 is a flowchart illustrating a false color correction interpolation according to the present invention, FIG. 3 is a diagram illustrating a regional area setting for boundary extraction in a Bayer color filter array, and FIG. 4 is a diagram for obtaining a boundary value in each direction. 5 is a diagram illustrating an interpolation mask, FIG. 5 is a diagram illustrating an interpolation region at the time of detection of a boundary, FIG. 6 is a diagram illustrating pixels for reproducing a color except for the color of a center pixel, and FIG. It is a figure which shows the example reproduced by the pixel.

Hereinafter, the configuration and operation of the present invention will be described with reference to FIGS. 2 to 7.

In performing the interpolation of the image sensor, first, eight boundary masks consisting of three adjacent pixels among eight pixels around an arbitrary pixel are set, and then an interpolation area is set (S200).

That is, (a), (b), and (c) of FIG. 3 each consist of 5 * 5 unit pixels, and represent three types of area areas for calculating a boundary value in the Bayer color pattern. When set, the values of the pixels at the positions indicated by the circles in FIG. 3 are applied to the edge value calculation mask, that is, the boundary mask of FIG. 4, divided into eight types of masks for determining the image boundary. The boundary value is calculated (S201).

The mask of FIG. 4 is used to determine an upper boundary, a lower boundary, a left boundary, a right boundary, an upper left boundary, a lower right boundary, an upper right boundary, and a lower left boundary on the basis of the center pixel. The boundary value of these eight masks is obtained as the absolute value of the difference of the sum 'A' of the product of the boundary value of the peripheral pixel and the mask value and the difference of the value 'B' of the center pixel and the mask value.

The boundary is determined by comparing a predetermined mask value 'B' with the intensity value of each of the eight boundary masks and a value 'A' multiplied by the intensity value of the pixel.

'A' is the sum of three pixels constituting each boundary mask of FIG. 4 and their values multiplied by their mask values a1, a2, and a3, respectively, and the mask value 'b' of the center pixel is dynamically changed according to the center pixel value. Assign. The dynamic mask value 'b' is assigned to adjust the specific gravity of the threshold value based on the center pixel value.

Here, a1, a2, a3 may be larger or smaller than b, and each of a1, a2, a3 may be the same or different.

For example, if a1 = 1, a2 = 2, a3 = 1, and b = 4 in the boundary mask # 1 of FIG. 4, when the intensities of the corresponding pixels are the same, {A | (a1 * corresponding pixel intensities) Value) + (a2 * corresponding pixel intensity value) + (a3 * corresponding pixel intensity value)} = {B | (b * center pixel intensity value)} to satisfy the absolute value of the difference between 'A' and 'B' Since the value becomes '0', no boundary exists in the boundary mask # 1 because the threshold value is not exceeded.

The threshold value is compared with a threshold value previously input to determine whether each direction boundary is present (S203), and when the value is higher than the threshold value for all the boundaries, the center pixel is a bad pixel having no relation with the values of the surrounding pixels. Judgment and defective pixel correction interpolation are performed (S204).

If it is determined that the defective pixel is not determined, if the boundary value of one or more directions is determined as the boundary through the threshold comparison, it is determined as the actual boundary (S205). As the central pixel corresponds to the boundary in one to seven directions, false color prevention interpolation is performed to exclude pixels outside the boundary from the interpolation region (S206).

If the center pixel does not correspond to the boundary in any direction, interpolation is performed to take the average of the pixels closest to the center pixel (S207).

5A and 5B show examples of interpolation regions in detecting edges in the green center pixel (pattern), respectively.

That is, in FIG. 5A, when the absolute value of the difference between the 'B' value of the green center pixel and the 'A' value of the boundary mask # 7 of FIG. 4 exceeds the threshold, the boundary exists only at the upper right side. In this case, a false color correction interpolation is performed, which is a correction interpolation using an area except a hatched area outside the boundary.

In FIG. 5B, when the absolute value of the difference between the 'B' value of the green center pixel and the 'A' value of the boundary mask # 4 of FIG. 4 exceeds the threshold value, the boundary exists only on the right side. In this case, false color correction interpolation, which is a correction interpolation using an area except the shaded area outside the boundary, is performed.

5 (c) and 5 (d) show examples of interpolation areas when detecting a boundary in a red center pixel, respectively. FIGS. 5 (e) and 5 (f) show a blue center pixel. Examples of the interpolation region at the time of boundary detection of FIG. 5 are the same as those described above with reference to FIGS. 5A and 5B, and thus description thereof is omitted.

6A to 6F show examples of color reproduction of immediately adjacent pixels except colors of the center pixel, respectively, and when interpolating colors other than the color of the center pixel, respectively. The case as shown in FIG. 7 can be considered.

In FIG. 7A, when the boundary is formed as shown in the red interpolation process in the green center pixel, the center is interpolated with only the red pixel on the left side, so that the red component is uniformly reproduced around the boundary.

The same applies to the blue color in FIG. 7B. Diagonal directions can be implemented in the same way. 7C and 7D show the case of a red center pixel. The blue center pixel is the same as the red center pixel.

In the case of (c) of FIG. 7, blue color is interpolated to the red center pixel, and thus, in the conventional median filter method, the false color cannot be filtered out. In the present invention, the false color can be prevented by ignoring pixels outside the boundary. . In the case of FIG. 7D, the green color is compensated for the red pixel, and similarly, unlike the conventional method, a false color of the green component can be prevented. This process is performed for all center pixels of the sensor.

According to the present invention, when color interpolation is performed using a memory, an array consisting of intensity values of a center pixel and intensity values of pixels having the same color of surroundings is constituted, and a mask for detecting a boundary in each direction is applied to the array, By interpolating differently depending on the presence or absence of a border, the embodiment has been found to minimize the false color generated at the border of the image and to improve the quality of the image.

Although the technical idea of the present invention has been described in detail according to the above preferred embodiment, it should be noted that the above-described embodiment is for the purpose of description and not of limitation. In addition, those skilled in the art will understand that various embodiments are possible within the scope of the technical idea of the present invention.

For example, the above description for correcting a bad pixel is an example in which the image information processing unit of the image sensor as shown in FIG. 1 performs color interpolation of a center pixel using an arrangement as shown in FIG. 3. The data processing method of the image sensor can be appropriately modified and applied based on the contents of the present invention in all cases using the interpolation method using the row memory.

The present invention as described above, by using Bayer data to determine whether the current interpolation position is the boundary of the image, and based on the result to distinguish the interpolation area to minimize the false color reproduction, to improve the quality of the image characteristics of the image sensor There is an effect to improve.

1 is a block diagram schematically showing an image sensor;

2 is a flowchart illustrating a false color correction interpolation according to the present invention.

FIG. 3 illustrates regional area settings for boundary extraction in a Bayer color filter array. FIG.

4 is a diagram showing a boundary mask for obtaining a boundary value in each direction.

5 is a diagram illustrating an interpolation region at the time of boundary detection.

FIG. 6 is a diagram illustrating pixels for reproducing a color except a color of a center pixel; FIG.

FIG. 7 is a diagram illustrating an example reproduced by the pixel of FIG. 6; FIG.

Claims (7)

Setting eight boundary masks composed of three pixels adjacent to each other among eight pixels around an arbitrary pixel; Setting an interpolation region; Determining a boundary by comparing a value 'B' obtained by multiplying an intensity value of a center pixel of the interpolation area by a preset mask value, and a value 'A' obtained by multiplying a corresponding value of each of the eight boundary masks and an intensity value of the pixel. ; Determining the center pixel as a bad pixel and performing bad pixel correction interpolation as the center pixel corresponds to a boundary in all eight directions; Performing false color prevention interpolation as the center pixel corresponds to a boundary in one to seven directions; And Performing interpolation using an average as the center pixel does not correspond to a boundary in all directions Color interpolation method of the image sensor comprising a. The method of claim 1, The eight boundary masks, And an upper boundary, a lower boundary, a left boundary, a right boundary, an upper left boundary, a lower right boundary, an upper right boundary, and a lower left boundary based on the center pixel, respectively. The method of claim 2, In determining the boundary, The 'A' is an image sensor color interpolation method characterized in that the sum of the product of the three mask values constituting each boundary mask and the intensity value of the pixel. The method of claim 3, wherein And a mask value of the center pixel is dynamically assigned. The method of claim 3, wherein In determining the boundary, And determining that the center pixel has a boundary in a corresponding boundary mask as an absolute value of the difference between 'A' and 'B' is greater than a predetermined threshold value. The method of claim 1, In the step of performing the bad pixel correction interpolation, A color interpolation method of an image sensor, characterized in that interpolation is performed using pixel values of the surroundings without using its own pixel values. The method of claim 1, In the step of performing false color interpolation, Color interpolation method of the image sensor, characterized in that the interpolation is performed using an area except the area corresponding to the boundary.