KR20130021977A - Device for correcting a difference of gr and gb sensitivity in image sensor and method thereof - Google Patents

Abstract

PURPOSE: An offset compensation device and a compensation method thereof are provided to compensate for offset between a Gr pixel and a Gb pixel. CONSTITUTION: An offset compensation device generates an input image as a pixel image of a five by five window size(S110). The offset compensation device defines a surrounding GB pixel including the largest offset(S120). The offset compensation device determines a compensation pixel value of a center pixel(S130). When the center pixel does not exist in an edge area, the offset compensation device outputs a pixel value of the center pixel(S150,S160). When the center pixel does not exist in the edge area, the offset compensation device outputs the compensation pixel value of the center pixel(S170). [Reference numerals] (AA) Start; (BB) Yes; (CC) No; (DD) End; (S110) Inputting a 5 x 5 window; (S120) Defining a surrounding pixel with the largest offset between a Gr pixel and a Gb pixel; (S130) Calculating the compensated pixel value(P33') of a center pixel; (S140) Verifying an edge region; (S150) Is the center pixel included in the edge region?; (S160) P33'=P33(maintaining a non-compensated value); (S170) Outputting P33'

Description

Device for correcting a pixel offset of an image sensor and a method of correcting the same

The present invention relates to an apparatus for correcting an offset between pixels of an image sensor and a method of correcting the same, and more particularly, to an apparatus for correcting an offset between pixels and a method for correcting an offset (cross-talk) between Gr and Gb pixels among pixels. will be.

In the CMOS image sensor (CIS), an image signal processing (ISP) block is a block for correcting or processing an input image by using a predetermined mathematical operation algorithm, which is applied to a physical structure of a pixel. The main function is to compensate for the deterioration of image quality inevitably.

1 is a diagram illustrating a general RGB color filter structure.

Referring to FIG. 1, the ISP block basically requires a function of correcting an offset between a Gr pixel and a Gb pixel generated when using the general RGB color filter structure shown in FIG. 1. Omission or lack of this correction function in the ISP block results in mazy noise in the output image, resulting in deterioration of image quality. Therefore, there is a need for offset correction or cross-talk compensation.

In addition, demosaic (interpolation, or interpolation) function of the ISP block converts Bayer-type image data into completed RGB-type image data. The above-described offset correction between Gr and Gb pixels is demosaic. It is done before the function is executed and is located in the image processing step of the Bayer area.

FIG. 2 is a diagram illustrating a basic configuration of a pixel array having the structure of the RGB color filter illustrated in FIG. 1, and FIG. 3 is a pixel array structure extending the basic configuration of the pixel array illustrated in FIG. The pixel array structure of the Bayer region extended in resolution is shown. 4 is a diagram illustrating a 5 x 5 window in which pixel positions in a 5 x 5 window used for calculation are matched in a Pxy form.

Referring to FIG. 3, in the pixel array structure of the extended Bayer region, the first dotted line box 30 reads an input image stored in a line memory, and then performs a mathematical operation to correct an offset between pixels of Gr and Gb. It represents a 5 x 5 window that is the base of the, and the second dotted line box 32 inside the first dotted line box 30 represents a center pixel currently undergoing correction through a 5 x 5 operation. That is, the center pixel (Gr pixel) indicated by the second dotted line box 32 is corrected using the peripheral homogeneous pixels (Gr pixel group and Gb pixel group) within the 5 × 5 window.

An existing equation for obtaining an offset between a Gr pixel and a Gb pixel may be expressed by Equation 1 below.

Here, the variable represented by Pxy in Equation 1 represents a coordinate variable in which the pixel position in the 5 × 5 window used for the operation is matched in the Pxy form as shown in FIG. 4.

As shown in Equation 1 above, the difference between the average value of the neighboring Gr pixels and the average value of the neighboring Gb pixels with respect to the center pixel Gr or P33 is defined as an offset, and then the operation calculates an offset between the Gr and Gb pixels. do.

In Equation 2 above, when the result value (offset value) using Equation 1 has a positive value, it is determined that the Gb pixel has a lower value than the Gr pixel even when the same green color is expressed, and thus the offset value is determined. It shows the operation of subtracting from the Gr pixel to maintain the level similar to the Gb pixel. On the contrary, if the offset value has a negative value, it is determined that the Gb pixel has a higher value than the Gr pixel even if the same green color is expressed, and the offset process is added to the Gr pixel to maintain a level similar to the Gb pixel. To show.

FIG. 5 is a diagram illustrating an example of a dead pixel in a 5 x 5 window used for an offset operation.

Referring to FIG. 5, when no dead pixel or defect pixel exists in a 5 × 5 window used for offset correction, normal correction may be performed.

However, as shown in FIG. 5 in the homogeneous pixel group (Gr pixel group and Gb pixel group) in the 5 x 5 window used for offset correction, when a dead pixel exists, the error value of the dead pixel affects the calculation process. (E.g. grating noise due to dead pixel value transition), resulting in an abnormal offset value and incorrect correction.

In particular, one or more dead pixels are often generated in the current trend toward a shared pixel structure. Therefore, even if a dead pixel exists in the homogeneous pixel group, an offset (or crosstalk) correction technique can be used. Development is needed.

Accordingly, an object of the present invention is to provide an image sensor that can solve problems such as generation of lattice noise due to a correction error in an offset correction process when a dead pixel exists in a homogeneous pixel group during offset correction between a Gr pixel and a Gb pixel. It is to provide an offset correction device between pixels.

Another object of the present invention is to provide an offset correction method between pixels of an image sensor that can solve problems such as generation of grid noise due to a correction error in the offset correction process.

In order to achieve the above object, an apparatus for offset correction between pixels of an image sensor according to an exemplary embodiment of the present invention includes one of the Gr and Gb pixels in an input pixel image having a pixel structure having a Gr pixel and a Gb pixel. A pixel-to-pixel offset correction device of an image sensor that corrects an offset between the Gr and Gb pixels within a window of a predetermined size defined such that a pixel is located at the center, the pixel being the same as the center pixel but not the dead pixel in the window. The average pixel value of the extracted pixels of the same kind is calculated, the calculated average pixel value and the pixel value of the center pixel are calculated, and peripheral pixels having the largest offset value with respect to the center pixel are calculated according to the calculation result. A selected Gx pixel offset calculating unit, a pixel value of a peripheral pixel having a largest offset value with respect to the center pixel, and a pixel of the center pixel Operation to, include Gx pixel offset correction, and the corrected pixel value to calculate the correction pixel value of the central pixel unit compensation pixel output for controlling the output.

Here, the pixel-to-pixel offset correction apparatus of the image sensor further includes an edge region determination unit for determining an edge region of the input pixel image and determining whether the center pixel exists in the edge region, and the correction pixel output unit includes the edge When the center pixel exists in the edge area according to the determination result of the area determining unit, the pixel value of the center pixel before correction is output without outputting the correction pixel value of the center pixel, and the center pixel is displayed in the edge area. If it does not exist, the pixel value of the center pixel is corrected to a corrected pixel value, and then output.

The offset correction method between pixels of an image sensor according to another aspect of the present invention is an offset correction method between pixels of an image sensor having a pixel structure having a Gr pixel and a Gb pixel, and includes any one of the Gr pixel and the Gb pixel. Determining a window having a predetermined size set to the center pixel, dividing an externally input input pixel image into a size of the determined window, and the center of the divided input pixel image rather than a dead pixel Selecting the homogeneous pixels of the pixel, selecting one homogeneous pixel having the largest offset value with the center pixel among the homogeneous pixels, and the selected one homogeneous pixel through a predetermined correction algorithm Compute the pixel value of the center pixel by calculating the pixel value of And a step of.

According to the present invention, even when a dead pixel exists in the same pixel group around the center pixel, it is possible to correct the offset between the Gr pixel and the Gb pixel, thereby improving the image quality deterioration.

In addition, when the present invention is implemented in an image sensor, the manufacturing process may be simplified due to component reduction and PCB wiring simplification, as compared with a case of implementing a separate ASIC.

In addition, when the present invention is implemented in the image sensor, compared to the case of implementing in a separate ASIC, it brings a lightening effect corresponding to one IC.

In addition, when the present invention is implemented in the image sensor, compared with the case of implementing a separate ASIC, it brings about the effect of reducing parts by one IC.

In addition, when the present invention is implemented in the image sensor, compared to the case of implementing a separate ASIC, it is possible to reduce the cost of components and assembly due to one IC and reduced PCB complexity.

1 is a diagram illustrating a general RGB color filter structure.
FIG. 2 is a diagram illustrating an arrangement structure of a pixel array having a structure of an RGB color filter illustrated in FIG. 1.
FIG. 2 is a diagram illustrating a basic configuration of a pixel array having a structure of an RGB color filter shown in FIG. 1.
FIG. 3 is a pixel array structure in which the basic configuration of the pixel array shown in FIG. 2 is extended, and illustrates a pixel array structure of a Bayer region extended at a resolution of 640 × 480.
4 is a diagram illustrating a 5 x 5 window in which pixel positions in a 5 x 5 window used for calculation are matched in a Pxy form.
FIG. 5 is a diagram illustrating an example of a dead pixel in a 5 x 5 window used for an offset operation.
6 is a block diagram illustrating an apparatus for offset correction between pixels of an image sensor according to an exemplary embodiment.
FIG. 7 is a flowchart illustrating an offset correction method when a dead pixel exists in a homogeneous pixel group in a 5 × 5 window.

In one embodiment of the present invention, when a dead pixel exists in a homogeneous pixel group in a 5x5 window, which is a mathematical calculation base used for offset correction between a Gr pixel and a Gb pixel, calculation with normal homogeneous pixels except for the dead pixel. Disclosed is a method for offset correction between pixels of an image sensor that may occur in offset correction.

In addition, in the pixel-to-pixel offset correction method according to an embodiment of the present invention, a technical configuration for determining an edge area is added, and the offset correction for the edge area is bypassed during offset correction, thereby providing a contrast ratio. A method of maintaining sharpness is disclosed.

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

6 is a block diagram illustrating an apparatus for offset correction between pixels of an image sensor according to an exemplary embodiment.

Referring to FIG. 6, an apparatus 100 for inter pixel offset correction of an image sensor according to an exemplary embodiment of the present invention may include a region divider 110, a Gx pixel offset calculator 120, and a Gx pixel offset corrector 130. ), An edge region determination unit 140, and a correction pixel output unit 150. Here, Gx written in the Gx pixel offset calculating unit 110 and the Gx pixel offset correcting unit 120, respectively, means any one of Gr and Gb, and the subscript x represents an extended concept including Gr and Gb. In other words, it is applicable if the center pixel in the 5x5 window, which is a mathematical operation base used for offset correction, is a green pixel regardless of Gr and Gb.

The area divider 110 receives an input image image through a Bayer color filter array, and divides the input image image into a pixel image having a predetermined window size. In this case, it is assumed that a Gr pixel or a Gx pixel is geometrically positioned at the center of the pixel image divided into the windows, and that the Gr pixel is positioned to maintain consistency of the description. The window size is defined as a size based on a mathematical operation used for offset correction between the Gr pixel and the Gb pixel, and is not particularly limited, but is assumed to be 5 (number of horizontal pixels) x 5 (number of vertical pixels) window size. do.

The Gx pixel offset calculating unit 120 receives a pixel image divided into 5 x 5 windows from the region dividing unit 110, scans the same pixel of the Gr pixel, and scans the homogeneous scanned image within the 5 x 5 window. Each pixel value of the pixels is analyzed to select normal homogeneous pixels, not dead pixels, and to obtain pixel values of neighboring pixels having the largest offset. Here, the pixel value may be a gray value.

The Gx pixel offset corrector 130 calculates the corrected pixel value of the corrected center pixel by calculating the neighboring pixel value having the largest offset obtained by the Gx pixel offset calculator 120 and the center pixel.

The edge region discrimination unit 140 is configured to determine an edge region for preventing the sharpness from falling off when offset correction is made in the edge region. The description will be omitted.

The correction pixel output unit 150 receives the correction pixel value of the center pixel and performs a function of limiting the correction pixel value when the input correction pixel value is out of a preset range. That is, if the corrected center pixel is out of the expressible range, this function limits it. For example, the correction pixel output unit 150 may perform a clamp function to prevent an overflow of the correction pixel value of the corrected center pixel.

In addition, the correction pixel output unit 150 performs a function of bypassing the corrected center pixel with respect to the edge region determined by the edge region determination unit 140. That is, it is determined whether the corrected pixel value of the corrected center pixel is output by using information on whether the center pixel exists in the edge area received by the edge area determination unit 130.

Hereinafter, an offset correction method in a case where a dead pixel exists in a homogeneous pixel group in a 5x5 window when offset correction between a Gr pixel and a Gb pixel will be described in detail with reference to FIG. 7.

FIG. 7 is a flowchart illustrating an offset correction method when a dead pixel exists in a homogeneous pixel group in a 5 × 5 window. For better understanding of the description, reference is made to FIG. 6 together.

Referring to FIG. 7, first, a pixel image obtained by dividing an input image image obtained through a Bayer color filter array into a 5 × 5 window size is generated (S110).

Subsequently, the first homogeneous Gr pixel group that surrounds the center pixel (Gr pixel) within the 5x5 window through the same configuration as the Gx pixel offset calculator 120 of FIG. 6 is expressed by a SORT function as shown in Equation 3 below. Selection through the algorithm.

An intermediate pixel value is selected from the group of homogeneous pixels selected through the SORT function. This primarily ensures that dead pixels are not included in the offset operation. Here, the process of selecting the intermediate pixel value is performed twice. The primary selection process is the process of selecting the middle pixel value from the outermost homogeneous pixel group in the window based on the center pixel, and the secondary selection process is the middle pixel of the homogeneous pixel group adjacent to the center pixel in the window. The process of selecting a value. The method of selecting the primary intermediate pixel value excludes the maximum pixel value and the minimum pixel value among the eight pixel values of the selected homogeneous pixel group. This leaves six pixel values. Among the remaining six pixel values, the remaining four pixel values except for the maximum pixel value and the minimum pixel value are selected as intermediate pixel values.

After the calculation of Equation 3, an average of four selected intermediate pixel values P _Gr3rd , P _Gr4th , P _Gr5th and P _Gr6th is obtained by applying the calculation algorithm represented by Equation 4 below.

Subsequently, as a process of selecting the second intermediate pixel value, a homogeneous Gb pixel group adjacent to the Gr center pixel is selected through a calculation algorithm represented by Equation 5 below, and two intermediate Gb in the selected second homogeneous Gr pixel group. Pixel value (P _Gb2nd , P _Gb3nd ) is selected. By doing so, the dead pixels included in the homogeneous Gb pixel group adjacent to the Gr center pixel are not included in the offset calculation process.

Subsequently, by applying an operation algorithm represented by Equation 6 below, the pixel having the largest difference from the two intermediate Gb pixel values selected in Equation 5 and the surrounding Gr pixel average value Gr_avg obtained by Equation 4 above is obtained. The offset is defined as the pixel around the largest Gb (S120).

Subsequently, when the peripheral Gb pixel having the largest offset value is determined through the above calculation, the offset (or cross) is calculated through the calculation algorithm represented by Equation 7 below through the same configuration as the Gx pixel offset correction unit 130 of FIG. 5. The corrected pixel value P33 'of the center pixel whose torque is corrected is determined (S130).

Subsequently, when the correction pixel value P33 of the center pixel is determined through Equation 7, it is determined whether the center Gr pixel exists in the edge area through the same configuration as that of the edge area determination unit 140 of FIG. 6. S140). The determination result is transmitted to the same configuration as the correction pixel output unit 140 of FIG. 6.

Subsequently, when the center pixel exists in the edge region (S150), the pixel value of the center pixel before correction is output as it is without outputting the correction pixel value of the center pixel (S160). On the contrary, if the center pixel does not exist in the edge region, the corrected pixel value of the center pixel is output (S170).

As described above, in the present invention, when a dead pixel exists in a homogeneous pixel group in a window of a specific size used for offset correction between a Gr pixel and a Gb pixel, the present invention is offset by calculating with normal homogeneous pixels excluding the dead pixel. Prevent errors that can occur in calibration. In addition, by adding a function of discriminating the edge region, the offset correction for the edge region is bypassed during offset correction, thereby maintaining the sharpness of the portion having contrast.

On the other hand, those skilled in the art will understand that the present invention can be implemented in other specific forms without changing the technical spirit or essential features. It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive. The scope of the present invention is defined by the appended claims rather than the detailed description, and all changes or modifications derived from the scope of the claims and their equivalents should be construed as being included within the scope of the present invention.

Claims (9)

In the pixel-to-pixel offset correction method of an image sensor having a pixel structure having a Gr pixel and a Gb pixel,
Determining a window having a predetermined size in which any one of the Gr pixel and the Gb pixel is set as a center pixel;
Dividing an input pixel image input to the outside into a size of the determined window;
Selecting homogeneous pixels of the center pixel that are not dead pixels in the window of the divided input pixel image;
Selecting one homogeneous pixel having the largest offset value with the center pixel among the homogeneous pixels: and
Calculating a correction pixel value for the center pixel by calculating a pixel value of the selected pixel of the same type and the pixel value of the center pixel through a predetermined correction algorithm;
The offset correction method between the pixels of the image sensor comprising a.
The method of claim 1, wherein selecting homogeneous pixels of the center pixel comprises:
Extracting pixel values of the homogeneous pixels (hereinafter, referred to as peripheral Gr pixels) located around the center Gr pixel when the center pixel is a Gr pixel (hereinafter referred to as a center Gr pixel);
Selecting intermediate pixel values excluding a maximum pixel value and a minimum pixel value from the extracted pixel values; And
Selecting peripheral Gr pixels corresponding to the selected intermediate pixel values as peripheral Gr pixels of the center Gr pixel instead of the dead pixel;
The offset correction method between the pixels of the image sensor comprising a.
The method of claim 2, wherein after the selecting with the surrounding Gr pixels,
Extracting respective pixel values of other homogeneous pixels (hereinafter, peripheral Gb pixels) located around the center Gr pixel;
Selecting intermediate pixel values excluding a maximum pixel value and a minimum pixel value from the extracted pixel values; And
Selecting peripheral Gr pixels corresponding to the selected intermediate pixel values as peripheral Gb pixels of the center Gr pixel instead of the dead pixel;
The offset correction method between the pixels of the image sensor comprising a. 4. The method of claim 3, wherein the step of selecting one homogeneous pixel having the largest offset value with the center Gr pixel,
Calculating an average pixel value of the peripheral Gr pixels;
Calculating a difference value between each pixel value of the peripheral Gb pixels and the calculated average value;
Selecting a peripheral Gb pixel corresponding to the largest difference among the calculated difference values as a homogeneous pixel having the largest offset value with the Gr pixel.
An offset correction method between pixels of the image sensor, characterized in that it comprises a.
The method of claim 4, wherein the calculating of the correction pixel value comprises:
Calculating a correction pixel value for the center Gr pixel by performing a correction operation on the pixel value of the center Gr pixel and the pixel value of the peripheral Gb pixel having the largest offset value with the center Gr pixel through the correction operation algorithm. A method for offset correction between pixels of an image sensor.
The method of claim 1, wherein after calculating a correction pixel value for the center pixel:
Determining whether the center pixel is in an edge region;
And determining whether to correct the center pixel according to the determination result.
The method of claim 6, wherein when the center pixel is in the edge region,
The pixel value of the center pixel before correction is maintained without outputting the correction pixel value of the center pixel,
If the center pixel is not in the edge region,
And a pixel value of the center pixel is corrected to a correction pixel value.
Correcting an offset between the Gr and Gb pixels within a window of a predetermined size defined such that any one of the Gr and Gb pixels is centered in an input pixel image having a pixel structure having Gr and Gb pixels. In the pixel-to-pixel offset correction device of the image sensor,
Compute the average pixel value of the extracted pixels of the same type by extracting the pixels of the same kind as the center pixel, not the dead pixel in the window, and calculates the calculated average pixel value and the pixel value of the center pixel, A Gx pixel offset calculator configured to select a peripheral pixel having a largest offset value with respect to the center pixel according to the first pixel;
A Gx pixel offset correction unit calculating a correction pixel value of the center pixel by calculating a pixel value of a peripheral pixel having the largest offset value with respect to the center pixel and a pixel value of the center pixel; And
A correction pixel output unit which controls the output of the correction pixel value;
Inter-pixel offset correction device of the image sensor comprising a.
The display apparatus of claim 8, further comprising an edge region discrimination unit configured to determine an edge region of the input pixel image, and determine whether the center pixel exists in the edge region.
The correction pixel output unit,
When the center pixel exists in the edge area according to the determination result of the edge area determination unit, the pixel value of the center pixel before correction is output without outputting the correction pixel value of the center pixel, and the center pixel is the edge. And the pixel value of the center pixel is corrected to a corrected pixel value when the pixel value does not exist in the region.

Images