CN112019775A

CN112019775A - Bad point detection and correction method and device

Info

Publication number: CN112019775A
Application number: CN202010922827.3A
Authority: CN
Inventors: 宋博; 王勇; 温建新
Original assignee: Chengdu Light Collector Technology Co Ltd
Current assignee: Chengdu Light Collector Technology Co Ltd
Priority date: 2020-09-04
Filing date: 2020-09-04
Publication date: 2020-12-01
Anticipated expiration: 2040-09-04
Also published as: CN112019775B

Abstract

The invention provides a bad point detection and correction method and a bad point detection and correction device, wherein the method comprises the following steps: determining a predetermined pixel point in the image data; determining whether a predetermined pixel point is a dead pixel or a suspected dead pixel; when the preset pixel point is a dead point, executing a first correction operation on the pixel value of the preset pixel point; when the predetermined pixel point is suspected to be a dead pixel, judging the image area of the predetermined pixel point in the image data, when the predetermined pixel point is judged to be located in the flat area of the image data, determining that the predetermined pixel point is a dead pixel, executing a second correction operation on the pixel value of the predetermined pixel point, when the predetermined pixel point is judged to be located in the edge area of the image data, judging whether the predetermined pixel point is a dead pixel again, and when the predetermined pixel point is judged to be a dead pixel, executing the second correction operation on the pixel value of the predetermined pixel point. The method has the advantages of high detection accuracy and correction accuracy, low cost and capability of ensuring the edge information and the detail information of the image.

Description

Bad point detection and correction method and device

Technical Field

The invention relates to the field of image processing, in particular to a bad point detection and correction method and device.

Background

When the image sensor outputs image data, because the positions of some pixels of the image sensor are subject to photosensitive failure, or because errors occur in the process of converting optical signals, the pixel values of some pixels in the image data output by the image sensor are inaccurate or have large differences with the pixel values of surrounding pixels, so that image dead pixels occur, and the imaging quality of images is affected. Therefore, it is generally necessary to detect whether or not there is a dead pixel in the image data and correct the dead pixel when there is a dead pixel.

In the related art, methods for detecting and correcting a dead pixel of image data generally include the following two methods:

the first method, static dead pixel removing method, specifically includes: and detecting pixel points with abnormal pixel values in the image data output by the image sensor through Automatic Test Equipment (ATE), determining the pixel points as dead pixels, and storing the position coordinates of the dead pixels. The algorithm processing module (ISP) can determine the position of the dead pixel based on the dead pixel position coordinates stored in the ATE, and the pixel value of the dead pixel is calculated and recovered by using the pixel values of the pixel points around the dead pixel, so that the dead pixel is corrected.

The second method, dynamic dead pixel removing method, specifically includes: the method comprises the steps of analyzing pixels of image data collected by an image sensor in real time, judging whether an error value between a pixel value of each pixel point of the image data and a pixel value of a pixel point around the pixel point is larger than a preset value or not, determining the pixel point as a dead pixel when the error value between the pixel value of a certain pixel point and the pixel value of the pixel point around the pixel point is larger than the preset value, and calculating and recovering the pixel value of the dead pixel by using the pixel value of the pixel point around the dead pixel so as to correct the dead pixel.

However, in the first method, when the number of dead pixels of a certain image data is too large, a large storage area is occupied, and thus the cost is increased. Meanwhile, for a part of pixel points, the pixel values of the pixel points may only show abnormality in the states of high and low temperature or high gain, and when the ATE detects dead pixels, the abnormality does not exist, so that the ATE cannot accurately detect all the dead pixels in the image data, and the detection accuracy and the correction accuracy are low. For the second method, because the determination standard for the dead pixel is single, the normal pixel point is usually mistakenly determined as the dead pixel or the dead pixel is missed, and the detection accuracy and the correction accuracy are both low.

Disclosure of Invention

The invention aims to provide a bad point detection and correction method and a bad point detection and correction device, which are used for solving the technical problems of higher cost, low detection accuracy and low correction accuracy of a bad point detection and correction method in the related technology.

In order to solve the above technical problem, the present invention provides a bad point detection and correction method, including:

determining a predetermined pixel point in the image data, wherein the predetermined pixel point is a pixel point to be detected;

determining an n × m matrix in the image data by taking the predetermined pixel point as a central pixel point, selecting a plurality of first reference pixel points which surround the predetermined pixel point and have the same color as the predetermined pixel point from the n × m matrix, and determining a maximum pixel value, a second maximum pixel value, a minimum pixel value and a second minimum pixel value in pixel values of the plurality of first reference pixel points;

determining whether the predetermined pixel point is a dead pixel or a suspected dead pixel based on the size relationship between the pixel value of the predetermined pixel point and the maximum pixel value, the second maximum pixel value, the minimum pixel value and the second minimum pixel value; when the difference value between the pixel value of the preset pixel point and the maximum pixel value is larger than a first preset value, or when the difference value between the minimum pixel value and the pixel value of the preset pixel point is larger than a second preset value, determining the preset pixel point as a dead pixel; when the difference value between the pixel value of the preset pixel point and the maximum pixel value is smaller than or equal to a first preset value, and the difference value between the pixel value of the preset pixel point and a second large pixel value is larger than the first preset value, or when the difference value between the minimum pixel value and the pixel value of the preset pixel point is smaller than or equal to a second preset value, and the difference value between the second small pixel value and the pixel value of the preset pixel point is larger than the second preset value, the preset pixel point is determined to be a suspected dead pixel; and

when the preset pixel point is a dead point, executing a first correction operation on the pixel value of the preset pixel point;

when the predetermined pixel point is a suspected dead pixel, based on the predetermined pixel point and the first reference pixel point having the maximum pixel value, or, determining a second reference pixel point in the n x m matrix based on a predetermined pixel point and a first reference pixel point having a minimum pixel value, and judges the image area of the predetermined pixel point in the image data based on the second reference pixel point, when the predetermined pixel point is judged to be located in the flat area of the image data, determining the predetermined pixel point as a dead pixel, performing a second correction operation on pixel values of predetermined pixel points, when it is judged that the predetermined pixel points are located in an edge area of the image data, judging whether the predetermined pixel points are dead points based on the second reference pixel points, and when the preset pixel point is judged to be a dead pixel point, executing the second correction operation on the pixel value of the preset pixel point.

Optionally, no other pixel point with the same color as the predetermined pixel point is arranged between the connecting line of each first reference pixel point and the predetermined pixel point.

Optionally, the method for determining whether the predetermined pixel point is a dead pixel or a suspected dead pixel based on the magnitude relationship between the pixel value of the predetermined pixel point and the maximum pixel value, the second maximum pixel value, the minimum pixel value, and the second minimum pixel value includes:

step one, judging whether the difference value between the pixel value of the preset pixel point and the maximum pixel value is larger than a first preset value or not, when the difference value between the pixel value of the preset pixel point and the maximum pixel value is larger than the first preset value, determining that the preset pixel point is a white dead pixel, otherwise, executing step two;

step two, judging whether the difference value between the pixel value of the preset pixel point and the second large pixel value is larger than a first preset value or not, and when the difference value between the pixel value of the preset pixel point and the second large pixel value is larger than the first preset value, determining that the preset pixel point is a suspected white dead pixel; otherwise, executing the third step;

judging whether the difference value between the minimum pixel value and the pixel value of the preset pixel point is larger than a second preset value or not, and determining that the preset pixel point is a black dead pixel when the difference value between the minimum pixel value and the pixel value of the preset pixel point is larger than the second preset value; otherwise, executing the step four;

step four, judging whether the difference value between the second small pixel value and the pixel value of the preset pixel point is larger than a second preset value or not, when the difference value between the second small pixel value and the pixel value of the preset pixel point is larger than the second preset value, determining that the preset pixel point is a suspected black dead pixel, and otherwise, executing step five;

and step five, determining the preset pixel points as normal pixel points.

Optionally, the first correction operation includes:

removing first reference pixel points which are positioned in the same row with the preset pixel points from the plurality of first reference pixel points to obtain a first pixel group, and calculating the transverse gradient value of the first pixel group; removing first reference pixel points which are positioned in the same row with the preset pixel points from the plurality of first reference pixel points to obtain a second pixel group, and calculating the longitudinal gradient value of the second pixel group;

judging whether the longitudinal gradient value and the transverse gradient value meet a first preset condition, wherein the first preset condition comprises the following steps: the longitudinal gradient value is smaller than a third preset value and the longitudinal gradient value is smaller than the difference between the transverse gradient value and a fourth preset value; when the longitudinal gradient value and the transverse gradient value meet a first preset condition, if the preset pixel point is a black dead pixel, determining the minimum value of the pixel values of first reference pixel points which are positioned in the same column with the preset pixel point as the corrected pixel value of the preset pixel point and outputting the pixel value; if the preset pixel point is a white dead pixel, determining the maximum value in the pixel values of the first reference pixel point which is positioned in the same column with the preset pixel point as the corrected pixel value of the preset pixel point and outputting the corrected pixel value;

judging whether the longitudinal gradient value and the transverse gradient value meet a second preset condition, wherein the second preset condition comprises the following steps: the transverse gradient value is smaller than a third preset value and the transverse gradient value is smaller than the difference between the longitudinal gradient value and a fourth preset value; when the longitudinal gradient value and the transverse gradient value meet a second preset condition, if the preset pixel point is a black dead pixel, determining the minimum value in the pixel values of the first reference pixel points which are positioned in the same row with the preset pixel point as the corrected pixel value of the preset pixel point and outputting the pixel value; if the preset pixel point is a white dead pixel, determining the maximum value in the pixel values of the first reference pixel point which is positioned in the same line with the preset pixel point as the corrected pixel value of the preset pixel point and outputting the corrected pixel value;

when the transverse gradient value and the longitudinal gradient value do not meet the first preset condition and the second preset condition, if the preset pixel point is a black dead pixel, determining the minimum pixel value as the pixel value of the corrected preset pixel point and outputting the pixel value; and if the preset pixel point is a white dead pixel, determining the maximum pixel value as the pixel value of the corrected preset pixel point and outputting the pixel value.

Optionally, the method for calculating the lateral gradient value of the first pixel group includes:

determining a first reference pixel point positioned at one side of a line where the predetermined pixel point is positioned in the first pixel group as a first sub-pixel group, sequentially arranging the first reference pixel points in the first sub-pixel group along the direction of the pixel line, and calculating an absolute value after the pixel values of adjacent first reference pixel points in the sequentially arranged first sub-pixel group are differed;

determining a first reference pixel point positioned on the other side of the line where the preset pixel point is positioned in the first pixel group as a second sub-pixel group, sequentially arranging the first reference pixel points in the second sub-pixel group along the pixel line direction, and calculating an absolute value after the pixel values of the adjacent first reference pixel points in the sequentially arranged second sub-pixel group are differenced;

and calculating the average value of the absolute values corresponding to the first pixel group, and determining the average value of the absolute values corresponding to the first pixel group as the transverse gradient value of the first pixel group.

Optionally, the method for calculating the longitudinal gradient value of the second pixel group includes:

determining a first reference pixel point positioned at one side of the row where the predetermined pixel point is positioned in the second pixel group as a third sub-pixel group, sequentially arranging the first reference pixel points in the third sub-pixel group along the direction of the pixel row, and calculating an absolute value after the pixel values of the adjacent first reference pixel points in the sequentially arranged third sub-pixel group are differed;

determining a first reference pixel point positioned on the other side of the row where the predetermined pixel point is positioned in the second pixel group as a fourth sub-pixel group, sequentially arranging the first reference pixel points in the fourth sub-pixel group along the pixel row direction, and calculating an absolute value after the pixel values of the adjacent first reference pixel points in the sequentially arranged fourth sub-pixel group are differenced;

and calculating the average value of the absolute values corresponding to the second pixel group, and determining the average value of the absolute values corresponding to the second pixel group as the longitudinal gradient value of the second pixel group.

Optionally, the method for determining the second reference pixel point in the nxm matrix based on the predetermined pixel point and the first reference pixel point with the maximum pixel value, or based on the predetermined pixel point and the first reference pixel point with the minimum pixel value includes:

when the predetermined pixel point is a suspected white dead pixel, the second reference pixel point is a pixel point which is closest to the first reference pixel point with the maximum pixel value in the extension direction of a connecting line of the predetermined pixel point and the first reference pixel point with the maximum pixel value;

and when the predetermined pixel point is a suspected black dead pixel, the second reference pixel point is a pixel point which is closest to the first reference pixel point with the minimum pixel value in the extension direction of a connecting line of the predetermined pixel point and the first reference pixel point with the minimum pixel value.

Optionally, the method for determining the image region of the predetermined pixel point in the image data based on the second reference pixel point includes:

selecting a plurality of third reference pixel points which surround the preset pixel points and have the same color as the second reference pixel points from the n x m matrix; judging whether the second reference pixel point and the third reference pixel point meet a third preset condition, wherein the third preset condition comprises that: the absolute value of the difference value between the pixel value of the second reference pixel point and the pixel value of each third reference pixel point is smaller than a third preset value; and when the third preset condition is met, determining that the preset pixel point is located in a flat area of the image data, otherwise, determining that the preset pixel point is located in an edge area of the image data.

Optionally, the method for determining whether the predetermined pixel point is a dead point based on the second reference pixel point includes:

when the colors of the second reference pixel point, the third reference pixel point and the preset pixel point are the same, if the preset pixel point is a suspected white dead pixel, judging whether the difference value between the pixel value of the preset pixel point and the pixel value of the second reference pixel point is larger than a first preset value, and if the judgment result is 'yes', determining that the preset pixel point is the white dead pixel; if the preset pixel point is a suspected black dead pixel, judging whether the difference value between the pixel value of the second reference pixel point and the pixel value of the preset pixel point is larger than a second preset value, and if the judgment result is yes, determining that the preset pixel point is a black dead pixel;

and when the colors of the second reference pixel point, the third reference pixel point and the preset pixel point are different, determining the preset pixel point as a normal pixel point.

Optionally, the second correction operation includes: when the preset reference point is determined to be a white dead pixel, determining the second large pixel value as the pixel value of the corrected preset pixel point and outputting the pixel value; and when the preset reference point is determined to be the black dead pixel, determining the second small pixel value as the pixel value of the corrected preset pixel point and outputting the pixel value.

In addition, the present invention also provides a dead pixel detection and correction apparatus, the apparatus including:

the first determining module is used for determining a preset pixel point in the image data, wherein the preset pixel point is a pixel point to be detected;

a second determining module, configured to determine an n × m matrix in the image data by using the predetermined pixel point as a central pixel point, select a plurality of first reference pixel points that surround the predetermined pixel point and have the same color as the predetermined pixel point from the n × m matrix, and determine a maximum pixel value, a second maximum pixel value, a minimum pixel value, and a second minimum pixel value among pixel values of the plurality of first reference pixel points;

the judging module is used for determining whether the preset pixel point is a dead pixel or a suspected dead pixel based on the size relation between the pixel value of the preset pixel point and the maximum pixel value, the second maximum pixel value, the minimum pixel value and the second minimum pixel value; when the difference value between the pixel value of the preset pixel point and the maximum pixel value is larger than a first preset value, or when the difference value between the minimum pixel value and the pixel value of the preset pixel point is larger than a second preset value, determining the preset pixel point as a dead pixel; when the difference value between the pixel value of the preset pixel point and the maximum pixel value is smaller than or equal to a first preset value, and the difference value between the pixel value of the preset pixel point and a second large pixel value is larger than the first preset value, or when the difference value between the minimum pixel value and the pixel value of the preset pixel point is smaller than or equal to a second preset value, and the difference value between the second small pixel value and the pixel value of the preset pixel point is larger than the second preset value, the preset pixel point is determined to be a suspected dead pixel; and

the dead pixel correction module is used for executing first correction operation on the pixel value of the preset pixel point when the preset pixel point is a dead pixel;

a suspected dead pixel correction module for correcting the predetermined pixel point based on the predetermined pixel point and the first reference pixel point having the maximum pixel value when the predetermined pixel point is a suspected dead pixel, or, determining a second reference pixel point in the n x m matrix based on a predetermined pixel point and a first reference pixel point having a minimum pixel value, and judges the image area of the predetermined pixel point in the image data based on the second reference pixel point, when the predetermined pixel point is judged to be located in the flat area of the image data, determining the predetermined pixel point as a dead pixel, performing a second correction operation on pixel values of predetermined pixel points, when it is judged that the predetermined pixel points are located in an edge area of the image data, judging whether the predetermined pixel points are dead points based on the second reference pixel points, and when the preset pixel point is judged to be a dead pixel point, executing the second correction operation on the pixel value of the preset pixel point.

In summary, in the method and the apparatus for detecting and correcting a dead pixel according to the present invention, when determining whether a predetermined pixel is a dead pixel, it is determined whether the predetermined pixel is a dead pixel or a suspected dead pixel based on the pixel value of the first reference pixel of the same color around the predetermined pixel. When the preset pixel point is determined to be a dead point, executing a first correction operation; when the predetermined pixel point is determined to be a suspected dead pixel, whether the suspected dead pixel is a dead pixel is further judged again, and when the suspected dead pixel is determined to be a dead pixel, a second correction operation is executed. In the invention, by adding the step of judging suspected dead pixels, the judgment process is more precise, and the phenomenon of 'misjudging normal pixel points as dead pixels' or 'dead pixel missing judgment' can be avoided, so that the dead pixel detection accuracy can be improved, and the image edge information and the detail information can be ensured.

Meanwhile, when the first correction operation is executed, the preset pixel point is corrected based on the transverse gradient value and the longitudinal gradient value of the pixels around the preset pixel point, and the correction accuracy is high.

In addition, the coordinates of the dead pixels do not need to be stored, so that the storage area can be saved, and the cost is reduced.

Drawings

Fig. 1 is a schematic flow chart of a bad point detection and correction method according to an embodiment of the present invention;

fig. 2 is a schematic diagram of a dot matrix of image data in an RGB Bayer format according to an embodiment of the present invention;

fig. 3 is a schematic structural diagram of a 5 × 5 matrix selected by using a predetermined pixel point as a central pixel point when the predetermined pixel point is a G pixel according to an embodiment of the present invention;

fig. 4 is a schematic diagram illustrating a position of a first reference pixel selected in the 5 × 5 matrix shown in fig. 3 according to an embodiment of the present invention;

fig. 5 is a schematic structural diagram of a 5 × 5 matrix selected by using a predetermined pixel point as a central pixel point when the predetermined pixel point is a B pixel according to an embodiment of the present invention;

fig. 6 is a schematic diagram illustrating a position of a first reference pixel in the 5 × 5 matrix shown in fig. 5 according to an embodiment of the present invention;

fig. 7a to 7h are schematic diagrams illustrating a selected position of the second reference pixel point when the predetermined pixel point is a G pixel point according to an embodiment of the present invention;

fig. 8a to 8h are schematic diagrams illustrating a selected position of the second reference pixel point when the predetermined pixel point is the B pixel point according to the embodiment of the present invention;

fig. 9a to 9h are schematic diagrams illustrating a selected position of the third reference pixel point when the predetermined pixel point is a G pixel point according to an embodiment of the present invention;

fig. 10a to 10h are schematic diagrams illustrating a selected position of the third reference pixel when the predetermined pixel is the B pixel according to an embodiment of the present invention.

Detailed Description

The method and apparatus for bad point detection and correction according to the present invention will be described in detail with reference to the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent from the following description. It is to be noted that the drawings are in a very simplified form and are not to precise scale, which is merely for the purpose of facilitating and distinctly claiming the embodiments of the present invention.

Fig. 1 is a schematic flow chart of a bad point detection and correction method according to an embodiment of the present invention, and as shown in fig. 1, the bad point detection and correction method includes:

step 100, obtaining image data, and determining a predetermined pixel point in the image data, wherein the predetermined pixel point is a pixel point to be detected.

Step 200, determining an n × m matrix in the image data by taking the predetermined pixel point as a central pixel point, selecting a plurality of first reference pixel points which surround the predetermined pixel point and have the same color as the predetermined pixel point from the n × m matrix, and determining a maximum pixel value, a second maximum pixel value, a minimum pixel value and a second minimum pixel value in pixel values of the plurality of first reference pixel points.

In this embodiment, the connection line between each of the first reference pixel points and the predetermined pixel point selected in this embodiment does not have other pixel points having the same color as the predetermined pixel point.

In addition, in this embodiment, n and m are both odd numbers, and the values of n and m should satisfy the following conditions: at least one first reference pixel capable of surrounding the predetermined pixel by one circle exists in the n × m matrix. Further, the values of n and m may differ for different formats of image data, wherein n may for example be equal to 3 and m may for example be equal to 3 when the image data comprises only monochrome color pixels, e.g. the image data is in a pure Gray (Gray) mode or when the image data comprises YUV format; and when the image data includes a plurality of color pixels, for example, when the image data includes: in the RGB Bayer format, the RGBW format, the RGB IR format, or the RCCB format, n may be equal to 5, for example, and m may be equal to 5, for example.

In addition, in this embodiment, eight first reference pixels may be specifically selected in the n × m matrix.

Step 300, determining whether the predetermined pixel point is a dead pixel or a suspected dead pixel based on the magnitude relation between the pixel value of the predetermined pixel point and the maximum pixel value, the second maximum pixel value, the minimum pixel value and the second minimum pixel value. When the preset pixel point is determined to be a dead point, executing step 400; when the predetermined pixel point is determined to be a suspected dead pixel, executing step 500; when it is determined that the predetermined pixel point is not a dead pixel and is not a suspected dead pixel (i.e., when the predetermined pixel point is a normal pixel point), step 800 is performed.

When the difference value between the pixel value of the preset pixel point and the maximum pixel value is larger than a first preset value, or when the difference value between the minimum pixel value and the pixel value of the preset pixel point is larger than a second preset value, determining the preset pixel point as a dead pixel; and when the difference value between the pixel value of the preset pixel point and the maximum pixel value is smaller than or equal to a first preset value, and the difference value between the pixel value of the preset pixel point and a second large pixel value is larger than the first preset value, or when the difference value between the minimum pixel value and the pixel value of the preset pixel point is smaller than or equal to a second preset value, and the difference value between the second small pixel value and the pixel value of the preset pixel point is larger than the second preset value, determining the preset pixel point as a suspected dead pixel.

In this embodiment, the first preset value and the second preset value may be the same or different, and the first preset value and the second preset value may be preset in the dead-pixel detection and correction device by an operator, where the first preset value and the second preset value may be specifically determined based on a gain value of an image sensor (e.g., CIS), and when the gain value of the image sensor is larger, the first preset value and the second preset value are also larger. For example, when the gain value of the image sensor is one gain, the first preset value and the second preset value may both be 200.

Step 400, executing a first correction operation to correct the pixel value of the predetermined pixel point and output the corrected pixel value of the predetermined pixel point.

Step 500, determining a second reference pixel point in the n × m matrix based on a predetermined pixel point and a first reference pixel point with a maximum pixel value, or based on a predetermined pixel point and a first reference pixel point with a minimum pixel value, and determining an image area of the predetermined pixel point in the image data based on the second reference pixel point, and when the predetermined pixel point is determined to be located in an edge area of the image data, executing step 600; when the predetermined pixel point is determined to be located in the flat area, the predetermined pixel point is determined to be a dead pixel, and step 700 is executed.

Step 600, judging whether the predetermined pixel point is a dead pixel point based on the second reference pixel point, executing step 700 when the predetermined pixel point is judged to be the dead pixel point, otherwise, determining that the predetermined pixel point is a normal pixel point, and executing step 800.

Step 700, executing a second correction operation to correct the pixel value of the predetermined pixel point and output the corrected pixel value of the predetermined pixel point.

And 800, directly outputting the pixel value of the preset pixel point.

In summary, in the dead pixel detection and correction method provided by the present invention, when determining whether a predetermined pixel point is a dead pixel, it is determined whether the pixel point is a dead pixel or a suspected dead pixel based on the pixel value of a first reference pixel point of the same color around the predetermined pixel point. When the preset pixel point is determined to be a dead point, executing a first correction operation; when the predetermined pixel point is determined to be a suspected dead pixel, whether the suspected dead pixel is a dead pixel is further judged again, and when the suspected dead pixel is determined to be a dead pixel, a second correction operation is executed. In the invention, by adding the step of judging suspected dead pixels, the judgment process is more precise, and the phenomenon of 'misjudging normal pixel points as dead pixels' or 'dead pixel missing judgment' can be avoided, so that the dead pixel detection accuracy can be improved, and the image edge information and the detail information can be ensured.

The black level detection and correction method provided by the embodiment of the invention is further described in detail below. The black level detection and correction method is mainly exemplified by taking an example that the format of image data includes an RGB Bayer format, and the specific method is as follows:

in step 100, the predetermined pixel point is selected from the image data, wherein in view of a subsequent need to establish an n × m matrix with the predetermined pixel point as a central pixel point, the predetermined pixel point should be any pixel point except a first row of pixel points, a last row of pixel points, a first column of pixel points, and a last column of pixel points of the image data in order to ensure that the n × m matrix can be successfully established.

And, in step 200, determining an n × m matrix in the image data by taking the predetermined pixel point as a central pixel point, and selecting a plurality of first reference pixel points surrounding the predetermined pixel point and having the same color as the predetermined pixel point from the n × m matrix.

Fig. 2 is a schematic diagram of a dot matrix of image data in an RGB Bayer format according to an embodiment of the present invention, and as shown in fig. 2, the image data in the RGB Bayer format includes R pixels, G pixels, and B pixels, and the predetermined pixel point may be an R pixel, a G pixel, or a B pixel. And in this step, a 5 × 5 matrix may be selected from the image data by taking the predetermined pixel point as a central pixel point, and then a plurality of first reference pixel points which surround the predetermined pixel point by one circle and have the same color as the predetermined pixel point are selected from the 5 × 5 matrix, wherein no other pixel point having the same color as the predetermined pixel point is provided between a connection line between each first reference pixel point and the predetermined pixel point.

It should be noted that, for the image data in RGB Bayer format, the periphery of the predetermined pixel point has a plurality of surrounding rings formed by pixel points with the same color as the predetermined pixel point, specifically, as shown in fig. 2, when the predetermined pixel point is a G pixel, the periphery of the predetermined pixel point has a plurality of surrounding rings formed by the same color, and two surrounding rings are schematically identified in fig. 2, and are respectively a surrounding ring a and a surrounding ring B. Wherein the annular ring A is the annular ring at the innermost ring close to the predetermined pixel point. In this embodiment of the present invention, the selected first reference pixels are substantially the pixels that form the ring circle a, that is: and no other pixel point with the same color as the preset pixel point is arranged between the connecting line of each first reference pixel point and the preset pixel point. Similarly, when the predetermined pixel point is a non-G pixel, the selected first reference pixel points should also be the pixel points forming the innermost ring circle.

Further, it should be noted that, when the pixel is a G pixel or a non-G pixel, the position of the first reference pixel selected in the n × m matrix is also different.

Specifically, fig. 3 is a schematic structural diagram of a 5 × 5 matrix selected by using a predetermined pixel point as a center pixel point when the predetermined pixel point is a G pixel according to an embodiment of the present invention, and fig. 4 is a schematic positional diagram of a first reference pixel point selected in the 5 × 5 matrix shown in fig. 3 according to an embodiment of the present invention, where as shown in fig. 4, eight first reference pixel points, namely, first reference pixel points Nb 0-Nb 7, are determined in the 5 × 5 matrix selected by using the G pixel point as the center pixel point, where the first reference pixel points Nb 0-Nb 7 surround the predetermined pixel point G for one circle, and no other pixel points with the same color are provided between each first reference pixel point and the predetermined pixel point.

Fig. 5 is a schematic structural diagram of a 5 × 5 matrix selected by using a predetermined pixel point as a central pixel point when the predetermined pixel point is a B pixel according to an embodiment of the present invention, and fig. 6 is a schematic structural diagram of a position of a first reference pixel point in the 5 × 5 matrix shown in fig. 5 according to an embodiment of the present invention. As shown in fig. 6, eight first reference pixels Nb [0] -Nb [7] are determined in a 5 × 5 matrix selected by taking a B pixel as a central pixel, where the first reference pixels Nb [0] -Nb [7] surround the predetermined pixel B for one circle, and no other pixels with the same color are provided between each first reference pixel and the predetermined pixel.

And after the first reference pixel points Nb 0-Nb 7 are determined, the pixel values of all the first reference pixel points can be obtained, and the pixel values of all the first reference pixel points are sorted in size to determine the maximum pixel value, the second maximum pixel value, the minimum pixel value and the second minimum pixel value in the first reference pixel points.

Further, in step 300, it may be determined whether the predetermined pixel point is a dead pixel or a suspected dead pixel based on a size relationship between the pixel value of the predetermined pixel point and the maximum pixel value, the second maximum pixel value, the minimum pixel value, and the second minimum pixel value.

It should be noted that, in a specific embodiment, the dead pixels are generally divided into black dead pixels and white dead pixels, the white dead pixels are pixel points whose pixel values are much higher than those of surrounding pixels, and the black dead pixels are pixel points whose pixel values are much lower than those of the surrounding pixels. Based on this, in step 300, the method for determining whether the predetermined pixel point is a dead pixel or a suspected dead pixel mainly includes:

step one, judging whether the difference value between the pixel value of the preset pixel point and the maximum pixel value is larger than a first preset value or not, and when the difference value between the pixel value of the preset pixel point and the maximum pixel value is larger than the first preset value, indicating that the pixel value of the preset pixel point is far larger than the pixel values of surrounding pixels, determining that the preset pixel point is a white dead pixel; when the difference between the pixel value of the predetermined pixel point and the maximum pixel value is smaller than or equal to a first preset value, and at this time, a situation that the first reference pixel point with the maximum pixel value is a dead pixel may exist, the predetermined pixel point cannot be completely determined to be a normal pixel point, and a second step needs to be executed to further determine the predetermined pixel point.

Step two, judging whether the difference value between the pixel value of the preset pixel point and the second large pixel value is larger than a first preset value or not, and when the difference value between the pixel value of the preset pixel point and the second large pixel value is larger than the first preset value, determining that the preset pixel point is a suspected white dead pixel; otherwise, executing step three.

Judging whether the difference value between the minimum pixel value and the pixel value of the preset pixel point is larger than a second preset value or not, and when the difference value between the minimum pixel value and the pixel value of the preset pixel point is larger than the second preset value, indicating that the pixel value of the preset pixel point is far smaller than the pixel values of surrounding pixels, and determining that the preset pixel point is a black dead pixel; when the difference between the minimum pixel value and the pixel value of the predetermined pixel point is less than or equal to a second preset value, and at this time, a situation that the first reference pixel point with the minimum pixel value is a dead pixel may exist, the predetermined pixel point cannot be completely determined to be a normal pixel point, and step four needs to be executed to further determine the predetermined pixel point.

And step four, judging whether the difference value between the second small pixel value and the pixel value of the preset pixel point is larger than a second preset value, when the difference value between the second small pixel value and the pixel value of the preset pixel point is larger than the second preset value, determining that the preset pixel point is a suspected black dead pixel, otherwise, executing step five.

And step five, determining the preset pixel points as normal pixel points.

And when it is determined in step 300 that the predetermined pixel point is a white dead pixel or a black dead pixel, step 400 may be performed, and when it is determined that the predetermined pixel point is a suspected white dead pixel or a suspected black dead pixel, step 500 may be performed; when the predetermined pixel point is determined to be a normal pixel point, step 800 is executed.

Wherein the step 400 mainly comprises: and executing a first correction operation to correct the pixel value of the preset pixel point.

The first correction operation may mainly include the steps of:

the method comprises the steps of firstly, removing first reference pixel points which are positioned in the same line with the preset pixel points in the first reference pixel points to obtain a first pixel group, calculating the transverse gradient value of the first pixel group, removing first reference pixel points which are positioned in the same column with the preset pixel points in the first reference pixel points to obtain a second pixel group, and calculating the longitudinal gradient value of the second pixel group.

In this embodiment, the reason for removing the first reference pixel located in the same row as the predetermined pixel from the first reference pixels is specifically: since the pixels with the same color adjacent to each other in the pixel row direction are usually differentiated when calculating the horizontal gradient value, for example, the first pixels Nb 0 and Nb 7, the first pixels Nb 7 and Nb 6, etc. in fig. 6, however, since the selected first reference pixels are all the pixels surrounding the predetermined pixel, the predetermined pixel is regularly spaced from the pixel in the same row as the predetermined pixel, but not the adjacent pixels with the same color, for example, the first pixels Nb 1 and Nb 5 in fig. 6 may be referred to, and at this time, for the calculation accuracy of the horizontal gradient value, the first reference pixel in the same row as the predetermined pixel in the first reference pixels is removed when calculating the horizontal gradient value.

Similarly, the first reference pixel point in the same column as the predetermined pixel point in the first reference pixel points is also removed when the vertical gradient value is calculated, so as to ensure the calculation accuracy of the vertical gradient value.

Further, the method for calculating the lateral gradient value of the first pixel group mainly comprises the following steps:

the first sub-step is to determine a first reference pixel point located at one side of the line where the predetermined pixel point is located in the first pixel group as a first sub-pixel group, arrange the first reference pixel points in the first sub-pixel group in sequence along the pixel line direction, and calculate an absolute value after the pixel values of the adjacent first reference pixel points in the sequentially arranged first sub-pixel group are subtracted.

If the format of the image data is an RGB Bayer format, then when the predetermined pixel point is a G pixel, referring to fig. 4, a first reference pixel point Nb [0], a first reference pixel point Nb [7], and a first reference pixel point Nb [6] located at one side of a row where the predetermined pixel point is located may be determined as a first sub-pixel group, and then the first reference pixel point Nb [0], the first reference pixel point Nb [7], and the first reference pixel point Nb [6] in the first sub-pixel group are sequentially ordered along a pixel row direction, which may specifically be: the first reference pixel Nb [0], the first reference pixel Nb [7], and the first reference pixel Nb [6], or the first reference pixel Nb [6], the first reference pixel Nb [7], and the first reference pixel Nb [0 ]. Then, the pixel values of the adjacent first reference pixel points in the first sub-pixel group after sequential arrangement are subjected to difference, and then an absolute value is obtained, that is, the following is calculated: i Nb 0-Nb 7I,

|Nb[7]-Nb[6]|。

And, if the format of the image data is RGB Bayer format, when the predetermined pixel point is a non-G pixel, and if the predetermined pixel point is a B pixel, referring to fig. 6, similarly, a first reference pixel point Nb [0], a first reference pixel point Nb [7], and a first reference pixel point Nb [6] located at one side of a row where the predetermined pixel point is located may be determined as a first sub-pixel group, and then the first reference pixel points in the first sub-pixel group are sequentially ordered along the pixel row direction, and then the pixel values of the adjacent first reference pixel points in the sequentially arranged first sub-pixel group are subtracted to obtain an absolute value, that is, the following calculation is performed: i Nb 0-Nb 7 and I Nb 7-Nb 6.

And a second sub-step of determining a first reference pixel point positioned on the other side of the row where the predetermined pixel point is positioned in the first pixel group as a second sub-pixel group, sequentially arranging the first reference pixel points in the second sub-pixel group along the pixel row direction, and calculating an absolute value after the pixel values of the adjacent first reference pixel points in the sequentially arranged second sub-pixel group are differed.

If the format of the image data is an RGB Bayer format, when the predetermined pixel point is a G pixel, similarly, a first reference pixel point Nb 2, a first reference pixel point Nb 3, and a first reference pixel point Nb 4 located on the other side of the line where the predetermined pixel point is located may be determined as a second sub-pixel group, and then the first reference pixel point Nb 2, the first reference pixel point Nb 3, and the first reference pixel point Nb 4 in the second sub-pixel group are sequentially sorted along the pixel line direction, and an absolute value is obtained after a difference is made between pixel values of adjacent first reference pixel points in the sequentially arranged second sub-pixel group, that is, the following calculation is performed: i Nb 2-Nb 3 and I Nb 3-Nb 4.

And if the format of the image data is an RGB Bayer format, when the predetermined pixel point is a non-G pixel, assuming that the predetermined pixel point is a B pixel, referring to fig. 6, similarly, it can be calculated that: i Nb 2-Nb 3 and I Nb 3-Nb 4.

A third substep of calculating an average of the absolute values obtained for the first pixel group and determining the average of the absolute values obtained for the first pixel group as the lateral gradient value of the first pixel group.

That is, the lateral gradient value gradsH of the first pixel group (| Nb [0] -Nb [7] | + | Nb [7] -Nb [6] | + | Nb [2] -Nb [3] | + | Nb [3] -Nb [4] |) ÷ 4.

Similarly, the method for calculating the longitudinal gradient value of the second pixel group may include the following steps:

and a fourth substep of determining a first reference pixel point positioned at one side of the row of the predetermined pixel points in the second pixel group as a third sub-pixel group, sequentially arranging the first reference pixel points in the third sub-pixel group along the direction of the pixel row, and calculating an absolute value after subtracting adjacent first reference pixel points in the sequentially arranged third sub-pixel group.

If the format of the image data is an RGB Bayer format, then when the predetermined pixel is a G pixel, referring to fig. 4, the first reference pixel Nb [0], the first reference pixel Nb [1], and the first reference pixel Nb [2] located at the column side of the predetermined pixel may be determined as a third sub-pixel group, and then the first reference pixel Nb [0], the first reference pixel Nb [1], and the first reference pixel Nb [2] in the third sub-pixel group are sequentially ordered along the pixel column direction, which may specifically be: the first reference pixel Nb [0], the first reference pixel Nb [1], the first reference pixel Nb [2], or the first reference pixel Nb [2], the first reference pixel Nb [1], the first reference pixel Nb [0 ]. Then, the pixel values of the adjacent first reference pixel points in the sequentially arranged third sub-pixel group are subjected to difference to calculate an absolute value, that is, the following is calculated: i Nb 0-Nb 1I,

|Nb[1]-Nb[2]|。

And, if the format of the image data is RGB Bayer format, when the predetermined pixel is a non-G pixel, if the predetermined pixel is a B pixel, referring to fig. 6, similarly, the first reference pixel Nb 0, the first reference pixel Nb 1, and the first reference pixel Nb 2 on the side of the row where the predetermined pixel is located may be determined as a third sub-pixel group, and the first reference pixels in the third sub-pixel group may be sequentially ordered along the pixel row direction, and the pixel values of the adjacent first reference pixels in the sequentially arranged third sub-pixel group are differenced to obtain an absolute value to calculate | Nb 0-Nb 1 |, | Nb 1-Nb 2 |.

And a fifth sub-step of determining a first reference pixel point positioned on the other side of the row where the predetermined pixel point is positioned in the second pixel group as a fourth sub-pixel group, sequentially arranging the first reference pixel points in the fourth sub-pixel group along the pixel row direction, and calculating an absolute value after subtracting adjacent first reference pixel points in the sequentially arranged fourth sub-pixel group.

If the format of the image data is an RGB Bayer format, when the predetermined pixel point is a G pixel, similarly, a first reference pixel point Nb 4, a first reference pixel point Nb 5, and a first reference pixel point Nb 6 located on the other side of the row where the predetermined pixel point is located may be determined as a fourth sub-pixel group, and then the first reference pixel point Nb 4, the first reference pixel point Nb 5, and the first reference pixel point Nb 6 in the fourth sub-pixel group are sequentially sorted along the pixel row direction, and an absolute value is obtained after a difference is made between pixel values of adjacent first reference pixel points in the sequentially arranged fourth sub-pixel group, that is, the following calculation is performed: i Nb 4-Nb 5 and I Nb 5-Nb 6.

And if the format of the image data is RGB Bayer format, when the predetermined pixel point is a non-G pixel and the predetermined pixel point is a B pixel, referring to FIG. 6, similarly, | Nb 4-Nb 5 |, | Nb 5-Nb 6 |, can be calculated.

A sixth substep of calculating an average of the absolute values obtained for the second pixel group and determining the average of the absolute values obtained for the second pixel group as the longitudinal gradient value of the second pixel group.

That is, the longitudinal gradient value gradsH of the second pixel group (| Nb [0] -Nb [1] | + | Nb [1] -Nb [2] | + | Nb [4] -Nb [5] | + | Nb [5] -Nb [6] |) ÷ 4.

And secondly, judging whether the longitudinal gradient value gradsV and the transverse gradient value gradsH meet a first preset condition, wherein the first preset condition comprises: the longitudinal gradient value gradsV is smaller than a third preset value and the longitudinal gradient value gradsV is smaller than a difference between the transverse gradient value gradsH and a fourth preset value; when the longitudinal gradient value gradsV and the transverse gradient value gradsH satisfy a first preset condition, if it is determined in the above step 300 that the predetermined pixel point is a black dead pixel, determining a minimum value of pixel values of a first reference pixel point located in the same column as the predetermined pixel point as a pixel value of the predetermined pixel point; if it is determined in step 300 that the predetermined pixel point is a white dead point, the maximum value among the pixel values of the first reference pixel point located in the same column as the predetermined pixel point is determined as the pixel value of the predetermined pixel point.

The third preset value and the fourth preset value may be specifically preset in the dead-pixel detection and correction device by an operator, where the third preset value and the fourth preset value may be specifically determined based on a gain value of an image sensor (for example, CIS), and when the gain value of the image sensor is larger, the third preset value and the fourth preset value are also larger. For example, the first preset value EDGE _ TH and the second preset value EDGE _ DIFF _ TH may both be 100.

And when gradsV < EDGE _ TH and gradsV < gradsH-EDGE _ DIFF _ TH; it is stated that in the n × m matrix, the longitudinal gradient value is smaller and much smaller than the transverse gradient value, that is, the change of the pixel point in the n × m matrix in the longitudinal direction is smaller, so that the predetermined pixel point can be corrected based on the pixel value of the first reference pixel point located in the same column as the predetermined pixel point, and the correction accuracy is higher. The maximum value or the minimum value of the pixel values of the first reference pixel points located in the same column as the predetermined pixel point may be determined as the pixel value of the predetermined pixel point, or an average value of the pixel values of the first reference pixel points located in the same column as the predetermined pixel point may be determined as the pixel value of the predetermined pixel point. In the embodiment of the invention, the maximum value or the minimum value of the pixel value of the first reference pixel point which is positioned in the same column with the preset pixel point is mainly determined as the pixel value of the preset pixel point, so that the image blurring can be prevented.

For example, in the embodiment of the present invention, when the format of the image data is the RGB Bayer format, and gradsV < EDGE _ TH and gradsV < gradsH-EDGE _ DIFF _ TH;

if the predetermined pixel point is determined to be a black dead point in step 300, the corrected pixel value of the predetermined pixel point is min (Nb 3, Nb 7);

if the predetermined pixel is determined to be a white dead pixel in step 300, the pixel value of the corrected predetermined pixel is max (Nb 3, Nb 7).

Thirdly, judging whether the longitudinal gradient value gradsV and the transverse gradient value gradsH meet a second preset condition, wherein the second preset condition includes: the transverse gradient value gradsH is smaller than a third preset value EDGE _ TH and the transverse gradient value gradsH is smaller than the difference between the longitudinal gradient value gradsV and a fourth preset value EDGE _ DIFF _ TH; when the longitudinal gradient value gradsV and the transverse gradient value gradsH satisfy a second preset condition, if the predetermined pixel point is a black dead pixel, determining a minimum value of pixel values of first reference pixel points located in the same row as the predetermined pixel point as a pixel value of the predetermined pixel point; and if the preset pixel point is a white dead pixel, determining the maximum value in the pixel values of the first reference pixel point which is positioned in the same line with the preset pixel point as the pixel value of the preset pixel point.

Similarly, when gradsH < EDGE _ TH and gradsH < gradsV-EDGE _ DIFF _ TH; it is stated that in the n × m matrix, the horizontal gradient value is small and much smaller than the vertical gradient value, that is, the change of the pixel point in the n × m matrix in the horizontal direction is small, and thus, the predetermined pixel point can be corrected based on the pixel value of the first reference pixel point located in the same row as the predetermined pixel point, and the correction accuracy is high. The maximum value or the minimum value of the pixel values of the first reference pixel points located in the same row as the predetermined pixel point may be determined as the pixel value of the predetermined pixel point, or the average value of the pixel values of the first reference pixel points located in the same row as the predetermined pixel point may be determined as the pixel value of the predetermined pixel point. In the embodiment of the present invention, the maximum value or the minimum value of the pixel value of the first reference pixel point located in the same column as the predetermined pixel point is mainly determined as the pixel value of the predetermined pixel point, so that the image blur can be prevented.

For example, in the embodiment of the present invention, when the format of the image data is RGB Bayer format and gradsV < EDGE _ TH and gradsV < gradsH-EDGE _ DIFF _ TH;

if the predetermined pixel point is determined to be a black dead point in step 300, the corrected pixel value of the predetermined pixel point is min (Nb 1, Nb 5);

if the predetermined pixel is determined to be a black dead pixel in step 300, the pixel value of the corrected predetermined pixel is max (Nb 1, Nb 5).

Fourthly, when the transverse gradient value and the longitudinal gradient value do not meet the first preset condition and the second preset condition, the image data can be regarded as a high-noise image, and at the moment, if the preset pixel point is a black dead pixel, the minimum pixel value is determined as the pixel value of the corrected preset pixel point and is output; and if the preset pixel point is a white dead pixel, determining the maximum pixel value as the corrected pixel value of the preset pixel point and outputting the pixel value.

Further, when it is determined in step 300 that the predetermined pixel point is suspected to be a dead pixel, step 500 is executed. The step 500 mainly comprises the following steps: and determining a second reference pixel point in the n x m matrix based on a predetermined pixel point and the first reference pixel point with the maximum pixel value, or based on the predetermined pixel point and the first reference pixel point with the minimum pixel value, and judging the image area of the predetermined pixel point in the image data based on the second reference pixel point.

The method for determining the second reference pixel point in step 500 may specifically be:

when it is determined in step 300 that the predetermined pixel point is a suspected white dead pixel, the second reference pixel point is a pixel point closest to the first reference pixel point with the maximum pixel value in the extending direction of the connection line between the predetermined pixel point and the first reference pixel point with the maximum pixel value.

When it is determined in step 300 that the predetermined pixel point is a suspected black dead pixel, the second reference pixel point is a pixel point closest to the first reference pixel point having the minimum pixel value in the extending direction of the connection line between the predetermined pixel point and the first reference pixel point having the minimum pixel value.

Fig. 7a to 7h are schematic diagrams of a selected position of the second reference pixel when the predetermined pixel is a G pixel according to the embodiment of the present invention; fig. 8a to 8h are schematic diagrams illustrating a selected position of the second reference pixel when the predetermined pixel is the B pixel according to the embodiment of the present invention. As shown in fig. 7a to 7h and fig. 8a to 8h, the position of the first reference pixel having the minimum pixel value or the position of the first reference pixel having the maximum pixel value is located at an EP position in the drawing, the position of the second reference pixel is located at an R [0] position in the drawing, and the second reference pixel R [0] is always a pixel closest to the EP in the extending direction of the connection line between the EP and the center pixel no matter whether the center pixel (i.e., the predetermined pixel) in the 5 × 5 matrix is a G pixel or a non-G pixel.

In this step 500, the method for determining the image region of the predetermined pixel point in the image data based on the second reference pixel point mainly includes:

and selecting a plurality of third reference pixel points which surround the preset pixel points and have the same color as the second reference pixel points from the n x m matrix. Determining an absolute value of a difference value between the pixel value of the second reference pixel and the pixel value of each third reference pixel, and judging whether the second reference pixel and the third reference pixel meet a third preset condition, where the third preset condition includes: the absolute value of the difference value between the pixel value of the second reference pixel point and the pixel value of each third reference pixel point is smaller than a third preset value; when the third preset condition is met, determining that the preset pixel point is located in the flat area of the image data, executing step 600, otherwise, determining that the preset pixel point is located in the edge area of the image data, and executing step 700.

Among them, it is recognized that the pixel values of pixels located in a flat area of image data are less different, and the pixel values of pixels located in an edge area of image data are more different. Based on this, when the second reference pixel point and the third pixel point satisfy a third preset condition, that is, when the absolute value of the difference between the pixel value of the second reference pixel point and the pixel value of each third reference pixel point is smaller than a third preset value, it may be determined that the pixel value of the second reference pixel point and the pixel value of each third reference pixel point are smaller, and it may be determined that the second reference pixel point and the third reference pixel point are located in a flat area of the image data, where in view of the fact that the third reference pixel point surrounds the predetermined pixel point, the predetermined pixel point should also be located in the flat area of the image data.

And when the second reference pixel point and the third pixel point do not satisfy a third preset condition, the difference between the pixel value of the second reference pixel point and the pixel value of at least one third reference pixel point is considered to be large, and then the second reference pixel point and the third reference pixel point are considered to be located in the edge area of the image data, and similarly, the predetermined pixel point is also located in the edge area of the image data.

It should be noted that when it is determined in this step 500 that the predetermined pixel point is located in the flat region of the image data, it may be determined that the predetermined pixel point is a dead pixel. Specifically, before step 500 is executed, it is determined that the predetermined pixel point is a suspected dead pixel (specifically, refer to step 300), that is, it is determined that a difference between a pixel value of the predetermined pixel point and a second large pixel value is greater than a first preset value, or a difference between a second small pixel value and a pixel value of the predetermined pixel point is greater than a second preset value. In other words, it has been determined that the difference between the pixel values of the predetermined pixel point and the pixel points around the predetermined pixel point is large. At this time, if it is determined in step 500 that the predetermined pixel point is located in the flat region of the image data, it may be concluded that: the preset pixel point is located in a flat area of the image data, but the difference between the pixel values of the preset pixel point and the pixel points around the preset pixel point is large. However, since the difference between the pixel values of the flat region is small under normal conditions, the predetermined pixel point can be determined to be a dead pixel, and therefore step 700 can be performed to correct the expected pixel point.

And, when it is determined in this step 500 that the predetermined pixel point is located in the edge area of the image data, it cannot be determined accurately that the predetermined pixel point is a dead pixel. Specifically, before step 500 is executed, it is determined that the predetermined pixel point is a suspected defective point (specifically, refer to step 300), that is, it is determined that the difference between the pixel values of the predetermined pixel point and the pixel points around the predetermined pixel point is large. At this time, if it is determined in step 500 that the predetermined pixel point is located in the edge region of the image data, it may be concluded that: the preset pixel points are located in the edge area of the image data, and the difference between the pixel values of the preset pixel points and the pixel values of the pixel points around the preset pixel points is large. However, in view of the fact that under normal conditions, the pixel values of the pixels in the edge region are different greatly, it cannot be determined that the predetermined pixel point is a dead pixel, and step 600 is performed to further determine the dead pixel.

As an example, in this step 500, three third reference pixels may be specifically determined, and fig. 9a to 9h are schematic diagrams of a selection position of the third reference pixel when the predetermined pixel is a G pixel according to an embodiment of the present invention. Fig. 10a to 10h are schematic diagrams illustrating a selected position of the third reference pixel when the predetermined pixel is the B pixel according to an embodiment of the present invention. As shown in FIGS. 9a-9h and FIGS. 10a-10h, the three selected third reference pixels are R1-R3, and the three third reference pixels R1-R3 have the same color as the second reference pixel R0 and surround the predetermined pixel.

And calculating an absolute value of a difference between the second reference pixel point and each third reference pixel point, wherein tmpTh1 ═ R [0] -R [1] |; tmpTh2 ═ R [0] -R [2] |; tmpTh3 ═ R0-R3 |.

When tmpTh1 < EDGE _ TH, tmpTh2 < EDGE _ TH, and tmpTh3 < EDGE _ TH are simultaneously established, it is determined that the predetermined pixel point is located in a flat region of the image data, and then step 700 may be performed. Otherwise, it is determined that the predetermined pixel point is located in the edge region of the image data, step 600 may be performed.

Step 600, judging whether the predetermined pixel point is a dead pixel point based on the second reference pixel point, executing step 700 when the predetermined pixel point is judged to be the dead pixel point, and executing step 800 when the predetermined pixel point is determined to be a normal pixel point.

The method for judging the predetermined pixel point as the dead point based on the second reference pixel point may include:

when the colors of the second reference pixel point, the third reference pixel point and the predetermined pixel point are the same (that is, corresponding to the situations of fig. 9a-9 d and fig. 10a-10 d), if the predetermined pixel point is determined to be a suspected white dead pixel in step 300, determining whether a difference between a pixel value of the predetermined pixel point and a pixel value of the second reference pixel point is greater than a first preset value, when the determination result is "yes", determining that the predetermined pixel point is a white dead pixel, otherwise, determining that the predetermined pixel point is a normal pixel point; if it is determined in step 300 that the predetermined pixel point is a suspected black dead pixel, determining whether a difference between a pixel value of the second reference pixel point and a pixel value of the predetermined pixel point is greater than a second preset value, if yes, determining that the predetermined pixel point is a black dead pixel, otherwise, determining that the predetermined pixel point is a normal pixel point.

When the colors of the second reference pixel point, the third reference pixel point and the predetermined pixel point are different (i.e. corresponding to the situations of fig. 9 e-9 h and fig. 10 e-10 h), it is determined that the predetermined pixel point is a normal pixel point.

Step 700, performing a second correction operation to correct the pixel value of the predetermined pixel point.

The second correction operation may specifically include: when the preset pixel point is determined to be a white dead pixel, determining the second large pixel value as the pixel value of the corrected preset pixel point; and when the preset reference point is determined to be the black dead pixel, determining the second small pixel value as the pixel value of the corrected preset pixel point.

And step 800, directly outputting the pixel value of the preset pixel point without executing correction operation.

It should be noted that, in the present invention, the bad point detection and correction method is described only by taking the image data as the RGB Bayer format as an example, but when the image data is the image data in other formats, for example, when the image data is in YUV format, RGBW format, RGB IR format, RCCB format, or the like, the bad point detection and correction method provided in the embodiment of the present invention is also applicable, and the bad point detection and correction method for the image data in other formats is similar to the bad point detection and correction method for the image data in the RGB Bayer format, and the embodiment of the present invention is not described herein again.

Further, the present invention provides a dead point detection correction apparatus for performing the dead point detection correction method shown in fig. 1, the apparatus including:

Optionally, no other pixel point with the same color as the predetermined pixel point is arranged between the connecting line of each first reference pixel point and the predetermined pixel point. Optionally, the dead pixel detection and correction device may include an algorithm processing (ISP) module, and the ISP module may be connected to the image sensor and configured to receive image data output by the image sensor.

In summary, the dead pixel detection and correction method and device provided by the invention have high dead pixel detection accuracy and correction accuracy, can ensure image edge information and detail information, and have low cost.

The above description is only for the purpose of describing the preferred embodiments of the present invention, and is not intended to limit the scope of the present invention, and any variations and modifications made by those skilled in the art based on the above disclosure are within the scope of the appended claims.

Claims

1. A method for dead pixel detection correction, the method comprising:

2. The method of claim 1, wherein each of the first reference pixels is connected to a predetermined pixel without any other pixel having a color same as that of the predetermined pixel.

3. The method of dead pixel detection and correction according to claim 1, wherein the method of determining whether the predetermined pixel point is a dead pixel or is a suspected dead pixel based on a magnitude relationship between the pixel value of the predetermined pixel point and the maximum pixel value, the second maximum pixel value, the minimum pixel value, and the second minimum pixel value comprises:

and step five, determining the preset pixel points as normal pixel points.

4. The dead-spot detection correction method of claim 3, wherein the first correction operation comprises:

5. The method of dead pixel detection correction of claim 4, wherein the method of calculating the lateral gradient values for the first pixel group comprises:

6. The method of dead pixel detection correction of claim 4, wherein the method of calculating the vertical gradient values of the second pixel group comprises:

7. The method of dead pixel detection correction of claim 3 wherein the method of determining a second reference pixel in said nxm matrix based on a predetermined pixel and a first reference pixel having a maximum pixel value or based on a predetermined pixel and a first reference pixel having a minimum pixel value comprises:

8. The method of dead pixel detection and correction according to claim 7, wherein the method of determining the image area in the image data to which the predetermined pixel belongs based on the second reference pixel comprises:

9. The dead pixel detection and correction method of claim 7, wherein the method of determining whether a predetermined pixel point is a dead pixel based on the second reference pixel point comprises:

10. The dead-spot detection correction method of claim 3, wherein the second correction operation comprises: when the preset reference point is determined to be a white dead pixel, determining the second large pixel value as the pixel value of the corrected preset pixel point and outputting the pixel value; and when the preset reference point is determined to be the black dead pixel, determining the second small pixel value as the pixel value of the corrected preset pixel point and outputting the pixel value.

11. A dead pixel detection correction apparatus, characterized in that the apparatus comprises: