CN105338342A - Image dead pixel detection method and device - Google Patents

Abstract

The invention relates to the image processing field, and particularly relates to an image dead pixel detection method and device, so as to solve the problem that when a rank dead pixel happens to a manufactured image sensor, the existing image dead pixel detection technology cannot detect the rank dead pixel problem. Through pixel values of a to-be-detected pixel and the same color pixel of the to-be-detected pixel in a detection window and a pixel difference value between the to-be-detected pixel and the same color pixel, whether a suspected rank dead pixel exits in the detection window is determined; and further according to the pixel difference value between the to-be-detected pixel and each same color pixel in the detection window, whether the to-be-detected pixel is the rank dead pixel is determined, and the fact that whether the to-be-detected pixel is the image dead pixel is judged according to single threshold information in the detection window can be avoided. Whether the rank dead pixel exists in the image sensor is detected, and rank dead pixel detection accuracy is improved.

Description

A kind of detection method of dead pixel points of images and device

Technical field

The present invention relates to image processing field, particularly relate to a kind of detection method and device of dead pixel points of images.

Background technology

Imageing sensor comprises a color filter array (colorfilterarray), i.e. Bayer filter, make the light of every bit collection on transducer by collecting a kind of light of wavelength after the filtering of Bayer filter, the light of every bit collection on transducer is red, blue or green by the light of the corresponding color passed through after filtering process.The light signal collected is converted into the signal of telecommunication by the photosensitive array in imageing sensor; Imageing sensor, again by the signal of telecommunication, is converted into digital signal, and the digital signal of conversion is exactly the pixel value of the some correspondence that on current sensor, each light gathers.The pixel that the array that imageing sensor is formed by the point of each light collection on transducer is corresponding forms original image corresponding to a width.The original image that imageing sensor obtains needs to carry out the relevant pre-process such as reduction noise and correction image bad point, to improve image displaying quality.Due in the manufacturing process of current transducer, if the array processes existing defects that the point that on the transducer in imageing sensor, each light gathers is formed, or light signal carries out occurring mistake in the process transformed, the information of some pixel on captured image can be caused wrong, cause the pixel value in image inaccurate, these defective pixels are dead pixel points of images (Badpixel).

As shown in Figure 1, for carrying out the schematic diagram of dead pixel points of images detection in prior art, wherein dotted portion is detection window, in detection window, the pixel of dash area is pixel to be detected, first determine the pixel value of pixel to be detected, and with pixel to be detected around the pixel value of same color pixel of same color; Then determine that the difference of the pixel value of the pixel value of pixel to be detected and the same color pixel of surrounding is whether in the pixel value threshold range set, if not in the threshold range of setting, then determine that this pixel to be detected is dead pixel points of images; Using with this dead pixel points of images around the pixel value of same color pixel to sue for peace the pixel value of averaging as after this dead pixel points of images corrects.

Traditional dead pixel points of images detection method, is applicable to the Hash image bad point of random distribution, i.e. normal image bad point, but cannot detects the dead pixel points of images of specific type, as ranks bad point; Ranks bad point refers to due to sensor process defect, a certain full line or permutation are all dead pixel points of images to cause the image collected to occur, these dead pixel points of images positions or entirely bright, entirely dark, or due in the process that light signal transforms, there is mistake in the conversion of certain a line or a certain row, causes the pixel of a row or column all to become dead pixel points of images; Adopting conventional method when detecting, when there is ranks bad point, because the value differences of the pixel of the same color of same a line or same row is little, cannot judge that this pixel to be detected is dead pixel points of images.

In sum, when occurring ranks bad point in detected image, existing dead pixel points of images detection technique cannot detect ranks bad point.

Summary of the invention

The object of this invention is to provide a kind of detection method of dead pixel points of images, to avoid judging that whether pixel to be detected is for dead pixel points of images according to threshold information single in detection window, the problem of ranks bad point cannot be detected, both achieved in detected image and whether there is ranks bad point, turn improved the accuracy of detected image bad point.

Embodiments provide a kind of detection method of dead pixel points of images, the method comprises:

For detected image, using the center of pixel to be detected as detection window, obtain the pixel value of described pixel to be detected and the same color pixel identical with described pixel color to be detected in detection window, and determine each pixel value difference with color pixel in described pixel to be detected and detection window respectively; Add up the quantity that described pixel value difference is greater than the same color pixel of the first pixel value difference threshold value, as the first quantitative value;

Determine described pixel to be detected in described detection window and with described pixel to be detected in same a line and the pixel value threshold range that all do not setting of the pixel value being positioned at the same color pixel on the right side of described pixel to be detected, and the absolute value of the pixel value difference of same color pixel on the right side of the described pixel to be detected determined and described pixel to be detected is all less than the absolute value threshold value of setting, then determine that described pixel to be detected is the doubtful row bad point of described detected image;

Determine described pixel to be detected in described detection window and with described pixel to be detected at same row and the pixel value threshold range that all do not setting of the pixel value being positioned at the same color pixel below described pixel to be detected, and the absolute value of the difference of the pixel value of same color pixel below the described pixel to be detected determined and described pixel to be detected is less than the absolute value threshold value of setting, then determine that described pixel to be detected is the doubtful row bad point of described detected image;

Determine that described first quantitative value is greater than the first amount threshold of setting, and determine that described pixel to be detected is the doubtful row bad point of described detected image, then determine that described pixel to be detected is the row bad point of described detected image; Determine that described first quantitative value is greater than the first amount threshold of setting, and determine that described pixel to be detected is the doubtful row bad point of described detected image, then determine that described pixel to be detected is the row bad point of described detected image.

By the pixel value of the same color pixel of pixel to be detected in detection window and pixel to be detected in above-described embodiment, and pixel to be detected with whether there is doubtful ranks bad point with in the pixel value difference determination detection window of color pixel; Determine whether this pixel to be detected is ranks bad point according to each pixel value difference with color pixel in pixel to be detected and detection window, avoid and judge whether pixel to be detected is dead pixel points of images according to threshold information single in detection window, both achieved in detected image and whether there is ranks bad point, turn improved the accuracy of detected image bad point.

In the embodiment of the present invention, the method also comprises:

Determine that described first quantitative value is not more than the first amount threshold of setting; And/or, determine that described pixel to be detected is not doubtful row bad point, neither doubtful row bad point;

Trigger and conventional bad point detection is carried out to described pixel to be detected.

Described in the embodiment of the present invention, conventional bad point detection is carried out to pixel to be detected, specifically comprises:

Add up each pixel value difference with color pixel in described pixel to be detected and detection window and be greater than the quantity of the same color pixel of the second pixel value difference threshold value, as the second quantitative value, if described second quantitative value is greater than the second amount threshold of setting, then determine that described pixel to be detected is conventional bad point, otherwise determine that described pixel to be detected is normal pixel, detection trigger window moves.

In above-described embodiment when determining pixel no enough row row bad point determination condition to be detected, conventional bad point detection being carried out to pixel to be detected, making the detection of pixel to be detected more accurate.

After determining in the embodiment of the present invention that described pixel to be detected is row bad point, then the longitudinal direction Grad of pixel to be detected described in described detection window, left diagonal Grad, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, after determining that described pixel to be detected is row bad point, then the horizontal direction Grad of pixel to be detected described in described detection window, left diagonal Grad, right diagonal Grad compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, determine that described pixel to be detected is row bad point, after being also row bad point, then the left diagonal Grad of pixel to be detected described in described detection window, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, determine that described pixel to be detected is not row bad point, neither after row bad point, then the horizontal direction Grad of pixel to be detected described in described detection window, longitudinal direction Grad, left diagonal Grad, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer.

In above-described embodiment after determining that pixel to be detected is dead pixel points of images, according to dead pixel points of images type, pixel to be detected is corrected, make the actual pixel value of the pixel value of the pixel to be detected after correction and pixel to be detected more close.

Determine in the embodiment of the present invention that the method for direction gradient value comprises:

To determine in described detection window with pixel to be detected in same a line and with a pair pixel value with color pixel of pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the horizontal direction Grad of described absolute value as pixel to be detected;

To determine in described detection window with pixel to be detected at same row and with a pair pixel value with color pixel of pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the longitudinal direction Grad of described absolute value as pixel to be detected;

To determine in described detection window on left diagonal and with a pair pixel value with color pixel of described pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the left diagonal Grad of described absolute value as pixel to be detected;

To determine in described detection window on right diagonal and with a pair pixel value with color pixel of described pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the right diagonal Grad of described absolute value as pixel to be detected.

Minimum direction gradient in the direction gradient value determined according to pixel to be detected in above-described embodiment is revised, and is conducive to retaining image edge details.

Embodiments provide a kind of checkout gear of dead pixel points of images, this device comprises:

Pixel value statistic unit, for for detected image, using the center of pixel to be detected as detection window, obtain the pixel value of described pixel to be detected and the same color pixel identical with described pixel color to be detected in detection window, and determine each pixel value difference with color pixel in described pixel to be detected and detection window respectively; Add up the quantity that described pixel value difference is greater than the same color pixel of the first pixel value difference threshold value, as the first quantitative value;

Doubtful row bad point determining unit, for determine described pixel to be detected in described detection window and with described pixel to be detected in same a line and the pixel value threshold range that all do not setting of the pixel value being positioned at the same color pixel on the right side of described pixel to be detected, and the absolute value of the pixel value difference of same color pixel on the right side of the described pixel to be detected determined and described pixel to be detected is all less than the absolute value threshold value of setting, then determine that described pixel to be detected is the doubtful row bad point of described detected image;

Doubtful row bad point determining unit, determine described pixel to be detected in described detection window and with described pixel to be detected at same row and the pixel value threshold range that all do not setting of the pixel value being positioned at the same color pixel below described pixel to be detected, and the absolute value of the difference of the pixel value of same color pixel below the described pixel to be detected determined and described pixel to be detected is less than the absolute value threshold value of setting, then determine that described pixel to be detected is the doubtful row bad point of described detected image;

Ranks bad point determining unit, for determining that described first quantitative value is greater than the first amount threshold of setting, and determines that described pixel to be detected is the doubtful row bad point of described detected image, then determines that described pixel to be detected is the row bad point of described detected image; Determine that described first quantitative value is greater than the first amount threshold of setting, and determine that described pixel to be detected is the doubtful row bad point of described detected image, then determine that described pixel to be detected is the row bad point of described detected image.

By the pixel value of the same color pixel of pixel to be detected in detection window and pixel to be detected in above-described embodiment, and pixel to be detected with whether there is doubtful ranks bad point with in the pixel value difference determination detection window of color pixel; Determine whether this pixel to be detected is ranks bad point according to each pixel value difference with color pixel in pixel to be detected and detection window, avoid and judge whether pixel to be detected is bad point according to threshold information single in detection window, both achieved in detected image transducer and whether there is ranks bad point, turn improved the accuracy of bad point in detected image.

The bad point of ranks described in embodiment of the present invention determining unit, also for:

Determine that described first quantitative value is not more than the first amount threshold of setting; And/or, determine that described pixel to be detected is not doubtful row bad point, neither doubtful row bad point; Trigger and conventional bad point detection is carried out to described pixel to be detected.

This device also comprises conventional dead pixel detecting unit, for:

Add up each pixel value difference with color pixel in described pixel to be detected and detection window and be greater than the quantity of the same color pixel of the second pixel value difference threshold value, as the second quantitative value, if described second quantitative value is greater than the second amount threshold of setting, then determine that described pixel to be detected is conventional bad point, otherwise determine that described pixel to be detected is normal pixel, detection trigger window moves.

In above-described embodiment when determining pixel no enough row row bad point determination condition to be detected, conventional bad point detection being carried out to pixel to be detected, making the detection of pixel to be detected more accurate.

In the embodiment of the present invention, this device also comprises bad point correcting unit, specifically for:

After determining that described pixel to be detected is row bad point, then the longitudinal direction Grad of pixel to be detected described in described detection window, left diagonal Grad, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, after determining that described pixel to be detected is row bad point, then the horizontal direction Grad of pixel to be detected described in described detection window, left diagonal Grad, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, determine that described pixel to be detected is row bad point, after being also row bad point, then the left diagonal Grad of pixel to be detected described in described detection window, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, determine that described pixel to be detected is not row bad point, neither after row bad point, then the horizontal direction Grad of pixel to be detected described in described detection window, longitudinal direction Grad, left diagonal Grad, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer.

In above-described embodiment after determining that pixel to be detected is dead pixel points of images, according to dead pixel points of images type, pixel to be detected is corrected, make the actual pixel value of the pixel value of the pixel to be detected after correction and pixel to be detected more close.

The correcting unit of bad point described in the embodiment of the present invention specifically for:

To determine in described detection window with pixel to be detected in same a line and with a pair pixel value with color pixel of pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the horizontal direction Grad of described absolute value as pixel to be detected;

To determine in described detection window with pixel to be detected at same row and with a pair pixel value with color pixel of pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the longitudinal direction Grad of described absolute value as pixel to be detected;

To determine in described detection window on left diagonal and with a pair pixel value with color pixel of described pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the left diagonal Grad of described absolute value as pixel to be detected;

To determine in described detection window on right diagonal and with a pair pixel value with color pixel of described pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the right diagonal Grad of described absolute value as pixel to be detected.

Minimum direction gradient in the direction gradient value determined according to pixel to be detected in above-described embodiment is revised, and is conducive to retaining image edge details.

Accompanying drawing explanation

Fig. 1 is the schematic diagram carrying out dead pixel points of images detection in prior art;

Fig. 2 is the schematic flow sheet of a kind of dead pixel points of images detection method in the embodiment of the present invention;

Fig. 3 be each transducer in imageing sensor after photosignal transforms, 4 kinds of arrangement mode schematic diagrames of the pixel color that each transducer is corresponding;

Fig. 4 is the schematic flow sheet of the detection method of a kind of ranks bad point in the embodiment of the present invention;

Fig. 5 be in the embodiment of the present invention to after pixel procession bad point detection to be detected, need the method flow schematic diagram carrying out conventional bad point detection;

Fig. 6 determines in the embodiment of the present invention that pixel to be detected is after ranks bad point, to the method flow schematic diagram that pixel to be detected corrects;

Fig. 7 is the schematic diagram of the checkout gear of a kind of dead pixel points of images in the embodiment of the present invention;

Fig. 8 is the schematic diagram of another kind of dead pixel points of images checkout gear in the embodiment of the present invention;

Fig. 9 is the schematic diagram in the direction of 4 acquisition direction gradient values in a detection window in the embodiment of the present invention;

Figure 10 a adopts the device in the embodiment of the present invention to carry out dead pixel points of images detection and the schematic diagram before correcting;

Figure 10 b adopts the device in the embodiment of the present invention to carry out dead pixel points of images detection and the schematic diagram after correcting.

Embodiment

The invention provides a kind of detection method of dead pixel points of images, by the pixel value of the same color pixel of pixel to be detected in detection window and pixel to be detected, and pixel to be detected with whether there is doubtful ranks bad point with in the pixel value difference determination detection window of color pixel; Determine whether this pixel to be detected is ranks bad point according to each pixel value difference with color pixel in pixel to be detected and detection window, avoid and judge whether pixel to be detected is dead pixel points of images according to threshold information single in detection window, both achieved in detected image transducer and whether there is ranks bad point, turn improved the accuracy of detected image bad point.

Below in conjunction with Figure of description, the present invention is further described.

As shown in Figure 2, be a kind of in embodiment of the present invention detection method of dead pixel points of images, the method comprises:

Step 201: for detected image, using the center of pixel to be detected as detection window, obtain the pixel value of pixel to be detected and the same color pixel identical with pixel color to be detected in detection window, and determine each pixel value difference with color pixel in pixel to be detected and detection window respectively; Statistical pixel difference is greater than the quantity of the same color pixel of the first pixel value difference threshold value, as the first quantitative value;

Step 202: determine pixel to be detected in detection window and with pixel to be detected in same a line and the pixel value threshold range that all do not setting of the pixel value being positioned at the same color pixel on the right side of pixel to be detected, and determine that the absolute value of the pixel value difference of the same color pixel on the right side of pixel to be detected and pixel to be detected is all less than the absolute value threshold value of setting, then determine that pixel to be detected is the doubtful row bad point of detected image;

Step 203: determine pixel to be detected in detection window and with pixel to be detected at same row and the pixel value threshold range that all do not setting of the pixel value being positioned at the same color pixel below pixel to be detected, and the absolute value of the difference of the pixel value of same color pixel below the pixel to be detected determined and pixel to be detected is less than the absolute value threshold value of setting, then determine that pixel to be detected is the doubtful row bad point of detected image;

Step 204: determine that the first quantitative value is greater than the first amount threshold of setting, and determine that pixel to be detected is the doubtful row bad point of detected image, then determine that pixel to be detected is the row bad point of detected image; Determine that the first quantitative value is greater than the first amount threshold of setting, and determine that pixel to be detected is the doubtful row bad point of detected image, then determine that pixel to be detected is the row bad point of detected image.

Wherein, in step 201, using the center of pixel to be detected as detection window, obtain the pixel value of described pixel to be detected and the same color pixel identical with described pixel color to be detected in detection window, specifically comprise: adopt Bayer filter, make every bit on sensor array can only collect a kind of light of wavelength, light signal is by being converted into the signal of telecommunication by light signal after the filtering of Bayer filter, after digitized processing, obtain the pixel of the some correspondence that each light gathers on imageing sensor.As shown in Figure 3, for each transducer in imageing sensor is after photosignal transforms, 4 kinds of arrangement modes of the pixel color that each transducer is corresponding.

Using the center of pixel to be detected as detection window, specifically comprise: the size of detection window is the window of (4m+1) * (4m+1) size, wherein m is positive integer.When the Partial Window of detection window is positioned at outside testing image picture, carry out automatic complement process.Wherein, when carrying out complement process, institute's complement picture is the capable and/or 2n row pixel of 2n, and n is positive integer at every turn.As when detection window top or low portion window are positioned at outside image to be detected, copy the capable pixel of 2n, and putting in order of the pixel of adjacent rows after copying is identical with the arrangement mode of the pixel of former image to be detected at every turn; And/or, when on the left of detection window or when right part window is positioned at outside image to be detected, copy 2n row pixel at every turn, and after copying adjacent two row pixels put in order identical with the arrangement mode of the pixel of former image to be detected; Image after making whole windows of detection window be arranged in complement process.

For the size of detection window for 5*5, when pixel to be detected is the first row first row of a two field picture, detection window is using the center of the pixel of the first row first row as detection window, detect that detection window upper portion is positioned at outside image to be detected, then the first row and the second row pixel are copied to the top of the first row pixel, and the second row pixel after copying contacts with former the first row pixel, form row duplicating image; Continue to detect detection window, detect that detection window left part is positioned at outside image to be detected, then the first row of row duplicating image and secondary series are copied to the left side of first row pixel, and the secondary series pixel after copying contacts with former first row pixel, form row duplicating image.

In step 201, through type (1) and formula (2) determine whether pixel to be detected is greater than the first pixel value difference threshold value with the pixel value difference with color pixel:

$\left\{\begin{array}{c}\mathrm{pBayer}\left[0\right]-\mathrm{pBayer}[2x*W+2y]>\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{diffval}\\ x^2+y^2\≠0;(-m\≤x\≤m,-m\≤y\≤m)\end{array}\right.---\left(1\right)$

Wherein, the pixel value that pBayer [0] is pixel to be detected, pBayer [2x*W+2y] is the pixel value of same color pixel identical with pixel color to be detected in detection window, and pBayer [0] > pBayer [2x*W+2y], reg_hvdpc_diffval is the first pixel value difference threshold value, x, y are integer, and W is the width of pixel place to be detected two field picture, and m is window control parameter to be detected.

$\left\{\begin{array}{c}\mathrm{pBayer}[2x*W+2y]-\mathrm{pBayer}\left[0\right]>\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{diffval}\\ x^2+y^2\≠0;(-m\≤x\≤m,-m\≤y\≤m)\end{array}\right.---\left(2\right)$

Wherein, the pixel value that pBayer [0] is pixel to be detected, pBayer [2x*W+2y] is the pixel value of same color pixel identical with pixel color to be detected in detection window, and pBayer [2x*W+2y] > pBayer [0], reg_hvdpc_diffval is the first pixel value difference threshold value, x, y are integer, and W is the width of pixel place to be detected two field picture, and m is window control parameter to be detected.

Add up the quantity that described pixel value difference is greater than the same color pixel of the first pixel value difference threshold value, (x when namely meeting formula (1) or formula (2), y) number of combinations, as the first quantitative value, judges whether the first quantitative value is greater than the first amount threshold of setting.

In step 202, to determine in detection window described pixel to be detected and with pixel to be detected on the right side of same a line with the pixel value of color pixel all not in the pixel value threshold range set, and determine that pixel to be detected and the right side with a line are all less than the absolute value threshold value of setting with the absolute value of the pixel value difference of color pixel, specifically comprise: for pixel to be detected in detection window and with pixel to be detected on the right side of same a line with color pixel i (0≤i≤2m, and i is even number) pixel value, if meet formula (3), the right side then representing pixel to be detected and go together with pixel to be detected is all very little with the brightness of color pixel, and the right side of going together with pixel to be detected is very little with the difference of the pixel value of color pixel and pixel to be detected:

$\left\{\begin{array}{c}\mathrm{pBayer}\left[i\right]<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{lowval}\\ \left|\mathrm{pBayer}\right[0]-\mathrm{pBayer}[i\left]\right|<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{sameval}\end{array}\right.---\left(3\right)$

Wherein, pBayer [i] is the pixel value of right side with color pixel being positioned at same row with pixel to be detected, reg_hvdpc_lowval is the lower limit of the pixel value threshold value of setting, and reg_hvdpc_sameval is the absolute value threshold value of setting, the pixel value that pBayer [0] is pixel to be detected.

Or, for pixel to be detected in detection window and with pixel to be detected on the right side of same a line with color pixel i (0≤i≤2m, and i is even number) pixel value, if meet formula (4), the right side then representing pixel to be detected and go together with pixel to be detected is all very large with the brightness of color pixel, and the right side of going together with pixel to be detected is very little with the difference of the pixel value of color pixel and pixel to be detected:

$\left\{\begin{array}{c}\mathrm{pBayer}\left[i\right]>\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{lowval}\\ \left|\mathrm{pBayer}\right[0]-\mathrm{pBayer}[i\left]\right|<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{sameval}\end{array}\right.---\left(4\right)$

Wherein, pBayer [i] is the pixel value of right side with color pixel being positioned at same row with pixel to be detected, reg_hvdpc_highval is the higher limit of the pixel value threshold value of setting, reg_hvdpc_sameval is the absolute value threshold value of setting, the pixel value that pBayer [0] is pixel to be detected.

When pixel to be detected and go together with pixel to be detected and the same color pixel be positioned on the right side of pixel to be detected all meets formula (3) or formula (4) time, then determine that this pixel to be detected is doubtful row bad point.

In step 203, to determine in detection window pixel to be detected and with pixel to be detected in the below of same row with the pixel value of color pixel all not in the pixel value threshold range set, and determine that the below of pixel to be detected and same row is less than the absolute value threshold value of setting with the absolute value of the difference of the pixel value of color pixel, specifically comprise: for pixel to be detected in detection window and with pixel to be detected in the below of same row with color pixel i (0≤i≤2m, and i is even number) pixel value, if meet formula (5), then represent pixel to be detected and all very little with the brightness of color pixel with the downside of pixel same column to be detected, and very little with the difference of the pixel value of color pixel and pixel to be detected with the downside of pixel same column to be detected:

$\left\{\begin{array}{c}\mathrm{pBayer}[i*W]<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{lowval}\\ \left|\mathrm{pBayer}\right[0]-\mathrm{pBayer}[i*W\left]\right|<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{sameval}\end{array}\right.---\left(5\right)$

Wherein, pBayer [i*W] is the pixel value of below with color pixel being positioned at same row with pixel to be detected, reg_hvdpc_lowval is the lower limit of the pixel value threshold value of setting, reg_hvdpc_sameval is the absolute value threshold value of setting, the pixel value that pBayer [0] is pixel to be detected, W is the width for pixel place to be detected two field picture.

Or, for pixel to be detected in detection window and with pixel to be detected in the below of same row with color pixel i (0≤i≤2m, its i is even number) pixel value, if meet formula (6), then represent pixel to be detected and all very large with the brightness of color pixel with the downside of pixel same column to be detected, and very little with the difference of the pixel value of color pixel and pixel to be detected with the downside of pixel same column to be detected:

$\left\{\begin{array}{c}\mathrm{pBayer}[i*W]>\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{lowval}\\ \left|\mathrm{pBayer}\right[0]-\mathrm{pBayer}[i*W\left]\right|<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{sameval}\end{array}\right.---\left(6\right)$

Wherein, pBayer [i*W] is the pixel value of below with color pixel being positioned at same row with pixel to be detected, reg_hvdpc_highval is the higher limit of the pixel value threshold value of setting, reg_hvdpc_sameval is the absolute value threshold value of setting, the pixel value that pBayer [0] is pixel to be detected.

When pixel to be detected and with pixel same column to be detected and the same color pixel be positioned on the downside of pixel to be detected all meets formula (5) or formula (6) time, then determine that this pixel to be detected is doubtful row bad point.

Specifically comprise in step 204: judge whether the first quantitative value is greater than the first amount threshold of setting, when determining that the first quantitative value is greater than the first amount threshold of setting, and determine that pixel to be detected is doubtful row dead pixel points of images, then determine that this pixel to be detected is row bad point; And/or, when determining that the first quantitative value is greater than the first amount threshold of setting, and determine that pixel to be detected is doubtful row bad point, then determine that this pixel to be detected is row bad point.

When determining that the first quantitative value is not more than the first amount threshold of setting, then carry out conventional bad point detection.When determining that the first quantitative value is greater than the first amount threshold of setting, but determine that pixel to be detected is not doubtful row bad point, and be not doubtful row bad point, then carry out conventional bad point detection.

Conventional bad point detection is carried out to pixel to be detected, specifically comprises: statistical pixel difference is greater than the quantity of the same color pixel of the second pixel value difference threshold value, as the second quantitative value, judge whether the second quantitative value is greater than the second amount threshold of setting; If the second quantitative value is greater than the second amount threshold of setting, then determine that pixel to be detected is conventional bad point, otherwise determine that pixel to be detected is normal pixel, detection trigger window moves.Wherein, the second pixel value difference threshold value is greater than the first pixel value difference threshold value under normal circumstances.

After determining that pixel to be detected is bad point, pixel to be detected is corrected, specifically comprise: after determining that pixel to be detected is conventional bad point, determine the horizontal direction Grad of pixel to be detected, longitudinal direction Grad, left diagonal Grad, right diagonal Grad, choosing the direction that in above-mentioned 4 directions, direction gradient value is minimum is reference direction; Determine that pixel to be detected is after row bad point, determine the longitudinal direction Grad of pixel to be detected, left diagonal Grad, right diagonal Grad, choosing the direction that in above-mentioned 3 directions, direction gradient value is minimum is reference direction; Determine that pixel to be detected is after row bad point, determine the horizontal direction Grad of pixel to be detected, left diagonal Grad, right diagonal Grad, choosing the direction that in above-mentioned 3 directions, direction gradient value is minimum is reference direction; Determining that pixel to be detected is row bad point, is again after row bad point, determines the left diagonal Grad of pixel to be detected, right diagonal Grad, and choosing the direction that in above-mentioned 2 directions, direction gradient value is minimum is reference direction.Average in the reference direction determined with the pixel value of same color pixel of pixel to be detected at a distance of 2p pixel unit, using the mean value that obtains as the pixel value after pixel correction to be detected, wherein p is positive integer.Can average in reference direction in detection window, as the pixel value after pixel correction to be detected with two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit; Also the pixel value of whole pixels homochromy with pixel to be detected in reference direction in detection window can be averaged, as the pixel value after pixel correction to be detected.

To average in reference direction in detection window with two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit, be example as the pixel value after pixel correction to be detected, if the reference direction that pixel to be detected is determined is horizontal direction, then the pixel value after pixel correction to be detected is such as formula shown in (7):

dpc_pBayer＝(pBayer[2p]+pBayer[-2p])/2(7)

Wherein, dpc_pBayer is the pixel value after pixel correction to be detected, pBayer [2p] and pBayer [-2p] for pixel to be detected in a lateral direction with pixel to be detected at a distance of two pixel values with color pixel of 2p pixel unit.

If the reference direction that pixel to be detected is determined is longitudinal direction, then the pixel value after pixel correction to be detected is such as formula shown in (8):

dpc_pBayer＝(pBayer[2p*W]+pBayer[-2p*W])/2(8)

Wherein, dpc_pBayer is the pixel value after pixel correction to be detected, pBayer [2p*W] and pBayer [-2p*W] on pixel longitudinal direction to be detected with pixel to be detected at a distance of two pixel values with color pixel of 2p pixel unit, W is the width of pixel place to be detected two field picture.

If the reference direction that pixel to be detected is determined is left diagonal, then the pixel value after pixel correction to be detected is such as formula shown in (9):

dpc_pBayer＝(pBayer[2p*(W+1)]+pBayer[-2p*(W+1)])/2(9)

Wherein, dpc_pBayer is the pixel value after pixel correction to be detected, pBayer [2p* (W+1)] and pBayer [-2p* (W+1)] in the left diagonal of pixel to be detected with pixel to be detected at a distance of two pixel values with color pixel of 2p pixel unit, W is the width of pixel place to be detected two field picture.

If the reference direction that pixel to be detected is determined is right diagonal, then the pixel value after pixel correction to be detected is such as formula shown in (10):

dpc_pBayer＝(pBayer[2p*(W-1)]+pBayer[-2p*(W-1)])/2(10)

Wherein, dpc_pBayer is the pixel value after pixel correction to be detected, pBayer [2p* (W-1)] and pBayer [-2p* (W-1)] in the right diagonal of pixel to be detected with pixel to be detected at a distance of two pixel values with color pixel of 2p pixel unit, W is the width of pixel place to be detected two field picture.

After completing pixel correction to be detected, write in pixel to be detected by the pixel value after correcting, the pixel value making next detection window obtain pixel to be detected in current detection window is the pixel value after correcting, and detection trigger window moves.

Determine that the method for the direction gradient value of pixel to be detected is as follows: such as formula to determine shown in (11) in detection window with pixel to be detected in same a line and with two absolute values with the pixel value difference of color pixel of pixel to be detected at a distance of 2p pixel unit, using the horizontal direction Grad of the absolute value of this pixel value difference as pixel to be detected;

hdir＝|pBayer[2p]-pBayer[-2p]|(11)

Wherein, hdir is the horizontal direction Grad of pixel to be detected, and pBayer [2p] and pBayer [-2p] is and two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit.

Such as formula to determine shown in (12) in detection window with pixel to be detected at same row and with two absolute values with the pixel value difference of color pixel of picture to be detected at a distance of 2p pixel unit, using the longitudinal direction Grad of the absolute value of this pixel value difference as pixel to be detected;

vdir＝|pBayer[2p*W]-pBayer[-2p*W]|(12)

Wherein, vdir is the longitudinal direction Grad of pixel to be detected, and pBayer [2p*W] and pBayer [-2p*W] is and two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit.

Such as formula to determine shown in (13) on the left diagonal of detection window and with two absolute values with the pixel value difference of color pixel of pixel to be detected at a distance of 2p pixel unit, using the left diagonal Grad of the absolute value of this pixel value difference as pixel to be detected;

ldir＝|pBayer[2p*(W+1)]-pBayer[-2p*(W+1)]|(13)

Wherein, ldir is the longitudinal direction Grad of pixel to be detected, and pBayer [2p* (W+1)] and pBayer [-2p* (W+1)] is and two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit.

Shown in (14), to determine on the right diagonal of detection window and with two absolute values with the pixel value difference of color pixel of pixel to be detected at a distance of 2p pixel unit, using the right diagonal Grad of the absolute value of this pixel value difference as pixel to be detected;

rdir＝|pBayer[2p*(W-1)]-pBayer[-2p*(W-1)]|(14)

Wherein, rdir is the longitudinal direction Grad of pixel to be detected, and pBayer [2p* (W-1)] and pBayer [-2p* (W-1)] is and two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit.

Preferably, in above-described embodiment calculated direction Grad and correct after pixel value time, when p gets 1, be one preferably execution mode, but p be other integer values, and also can meet the embodiment of the present invention require value, also the present invention for protect scope in.

In the embodiment of the present invention, detection window is according to from left to right, order from top to bottom moves, when determining doubtful row bad point and doubtful row bad point, go together with pixel to be detected in detection window, and the pixel be positioned on the left of pixel to be detected, and with pixel same column to be detected, and the pixel be positioned on the upside of pixel to be detected is all the pixel after correcting, or be not the pixel of bad point, when therefore pixel to be detected being detected in the embodiment of the present invention, checking and pixel to be detected are gone together, and the pixel be positioned on the right side of pixel to be detected, and with pixel same column to be detected, and the pixel be positioned on the downside of pixel to be detected, when detection window moves according to other orders, when pixel to be detected is detected, determine whether pixel to be detected is dead pixel points of images according to the pixel do not detected.

As shown in Figure 4, be a kind of in embodiment of the present invention detection method of ranks bad point, suppose that the size of window to be detected is 5*5 pixel, the method comprises:

Step 401: for detected image, using the center of pixel to be detected as detection window;

Step 402: judge whether detection window top or low portion window are positioned at outside image to be detected, if having, then performs step 403, otherwise performs step 404;

Step 403: copy 2 row pixels on detection window top or low portion window, and putting in order of the pixel of two row copied is identical with the arrangement mode of the pixel of former image to be detected;

Step 404: judge on the left of detection window or whether right part window is positioned at outside image to be detected, if having, then performs step 405, otherwise performs step 406;

Step 405: on the left of detection window or right part window copy 2 row pixels, and copy two row pixels put in order identical with the arrangement mode of the pixel of former image to be detected;

Step 406: the pixel value obtaining pixel to be detected and the same color pixel identical with pixel color to be detected in detection window;

Step 407: determine each pixel value difference with color pixel in pixel to be detected and detection window respectively;

Step 408: statistical pixel difference is greater than the quantity of the same color pixel of the first pixel value difference threshold value, as the first quantitative value;

Step 409: judge whether the first quantitative value is greater than the first amount threshold of setting, if be greater than, then performs step 410, otherwise performs step 412;

Step 410: judge the whether doubtful row bad point of pixel to be detected, if not, then perform step 411, otherwise perform step 413;

Step 411: judge the whether doubtful row bad point of pixel to be detected, if not, then perform step 412, otherwise perform step 414;

Step 412: conventional bad point detection is carried out to pixel to be detected, and process ends;

Step 413: determine that pixel to be detected is row bad point, row bad point mark is set to 1, and process ends;

Step 414: determine that pixel to be detected is row bad point, row bad point mark is set to 1, and process ends.

As shown in Figure 5, in the embodiment of the present invention to after pixel procession bad point detection to be detected, need the method flow schematic diagram carrying out conventional bad point detection, the method comprises:

Step 501: conventional bad point detection is carried out to pixel to be detected;

Step 502: statistical pixel difference is greater than the quantity of the same color pixel of the second pixel value difference threshold value, as the second quantitative value;

Step 503: judge whether the second quantitative value is greater than the second amount threshold of setting, if be not more than, performs step 504, otherwise performs step 505;

Step 504: determine that pixel to be detected is normal pixel, detection trigger window moves, and process ends;

Step 505: determine that pixel to be detected is conventional bad point;

Step 506: be reference direction by direction minimum for direction gradient value in the horizontal direction Grad of pixel to be detected, longitudinal direction Grad, left diagonal Grad, right diagonal Grad;

Step 507: average in the reference direction determined at a distance of 2 pixel values with color pixel of 2 pixel units with pixel to be detected, using the mean value that obtains as the pixel value after pixel correction to be detected;

Step 508: the pixel value after correcting is write pixel to be detected, and process ends.

As shown in Figure 6, after determining in the embodiment of the present invention that pixel to be detected is ranks bad point, to the method that pixel to be detected corrects, the method comprises:

Step 601: determine that pixel to be detected is ranks bad point;

Step 602: judge row bad point mark, and whether row bad point mark is set to 1 simultaneously, if so, then performs step 603, otherwise perform step 604;

Step 603: determine the left diagonal Grad of pixel to be detected, right diagonal Grad, the minimum direction of selected directions Grad is reference direction, and performs step 607;

Step 604: judge whether row bad point mark is set to 1, if so, then performs step 605, otherwise performs step 606;

Step 605: determine the longitudinal direction Grad of pixel to be detected, left diagonal Grad, right diagonal Grad, the minimum direction of selected directions Grad is reference direction, and performs step 607;

Step 606: determine the horizontal direction Grad of pixel to be detected, left diagonal Grad, right diagonal Grad, the minimum direction of selected directions Grad is reference direction;

Step 607: by the reference direction determined is averaged at a distance of 2 pixel values with color pixel of 2 pixel units with pixel to be detected, using the mean value that obtains as the pixel value after pixel correction to be detected;

Step 608: the pixel value after correcting is write pixel to be detected, and process ends.

Based on same inventive concept, a kind of checkout gear of dead pixel points of images is additionally provided in the embodiment of the present invention, the principle of dealing with problems due to this device is similar to the detection method of a kind of dead pixel points of images of the embodiment of the present invention, therefore the enforcement of this device see the enforcement of method, can repeat part and repeats no more.

As shown in Figure 7, be a kind of in embodiment of the present invention checkout gear of dead pixel points of images, this device comprises pixel value statistic unit 701, doubtful row bad point determining unit 702, doubtful row bad point determining unit 703, ranks bad point determining unit 704:

Pixel value statistics single 701, for for detected image, using the center of pixel to be detected as detection window, obtain the pixel value of pixel to be detected and the same color pixel identical with pixel color to be detected in detection window, and determine each pixel value difference with color pixel in pixel to be detected and detection window respectively; Statistical pixel difference is greater than the quantity of the same color pixel of the first pixel value difference threshold value, as the first quantitative value;

Doubtful row bad point determining unit 702, for determine pixel to be detected in detection window and with pixel to be detected in same a line and the pixel value threshold range that all do not setting of the pixel value being positioned at the same color pixel on the right side of pixel to be detected, and the absolute value of the pixel value difference of same color pixel on the right side of the pixel to be detected determined and pixel to be detected is all less than the absolute value threshold value of setting, then determine that pixel to be detected is the doubtful row bad point of detected image;

Doubtful row bad point determining unit 703, determine described pixel to be detected in detection window and with pixel to be detected at same row and the pixel value threshold range that all do not setting of the pixel value being positioned at the same color pixel below pixel to be detected, and the absolute value of the difference of the pixel value of same color pixel below the pixel to be detected determined and pixel to be detected is less than the absolute value threshold value of setting, then determine that pixel to be detected is the doubtful row bad point of detected image;

Ranks bad point determining unit 704, for determining that the first quantitative value is greater than the first amount threshold of setting, and determines that pixel to be detected is the doubtful row bad point of detected image, then determines that pixel to be detected is row bad point; Determine that the first quantitative value is greater than the first amount threshold of setting, and determine that pixel to be detected is the doubtful row bad point of detected image, then determine that pixel to be detected is the row bad point of detected image.

Wherein, pixel value statistic unit 701, specifically for: adopt Bayer filter, make each transducer can only collect a kind of light of wavelength, light signal, by light signal being converted into the signal of telecommunication after the filtering of Bayer filter, after digitized processing, obtains the pixel that each imageing sensor is corresponding.

The size of detection window is the window of (4m+1) * (4m+1) size, and wherein m is positive integer.When the Partial Window of detection window is positioned at outside testing image picture, carry out automatic complement process.Wherein, when carrying out complement process, institute's complement picture is the capable and/or 2n row pixel of 2n, and n is positive integer at every turn.As when detection window top or low portion window are positioned at outside image to be detected, copy the capable pixel of 2n, and putting in order of the pixel of adjacent rows after copying is identical with the arrangement mode of the pixel of former image to be detected at every turn; And/or, when on the left of detection window or when right part window is positioned at outside image to be detected, copy 2n row pixel at every turn, and after copying adjacent two row pixels put in order identical with the arrangement mode of the pixel of former image to be detected; Image after making whole windows of detection window be arranged in complement process.

Pixel value statistic unit 701, specifically for: through type (1) and formula (2) determine whether pixel to be detected is greater than the first pixel value difference threshold value with the pixel value difference with color pixel.

$\left\{\begin{array}{c}\mathrm{pBayer}\left[0\right]-\mathrm{pBayer}[2x*W+2y]>\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{diffval}\\ x^2+y^2\&NotEqual;0;(-m\&le;x\&le;m,-m\&le;y\&le;m)\end{array}\right.---\left(1\right)$

Wherein, the pixel value that pBayer [0] is pixel to be detected, pBayer [2x*W+2y] is the pixel value of same color pixel identical with pixel color to be detected in detection window, and pBayer [0] > pBayer [2x*W+2y], reg_hvdpc_diffval is the first pixel value difference threshold value, x, y are integer, and W is the width of pixel place to be detected two field picture, and m is window control parameter to be detected.

$\left\{\begin{array}{c}\mathrm{pBayer}[2x*W+2y]-\mathrm{pBayer}\left[0\right]>\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{diffval}\\ x^2+y^2\&NotEqual;0;(-m\&le;x\&le;m,-m\&le;y\&le;m)\end{array}\right.---\left(2\right)$

Wherein, the pixel value that pBayer [0] is pixel to be detected, pBayer [2x*W+2y] is the pixel value of same color pixel identical with pixel color to be detected in detection window, and pBayer [2x*W+2y] > pBayer [0], reg_hvdpc_diffval is the first pixel value difference threshold value, x, y are integer, and W is the width of pixel place to be detected two field picture, and m is window control parameter to be detected.

Add up the quantity that described pixel value difference is greater than the same color pixel of the first pixel value difference threshold value, as the first quantitative value, judge whether the first quantitative value is greater than the first amount threshold of setting.

Doubtful row bad point determining unit 702, specifically for: for pixel to be detected in detection window and with pixel to be detected on the right side of same a line with color pixel i (0≤i≤2m, and i is even number) pixel value, if meet formula (3), the right side then representing pixel to be detected and go together with pixel to be detected is all very little with the brightness of color pixel, and the right side of going together with pixel to be detected is very little with the difference of the pixel value of color pixel and pixel to be detected:

$\left\{\begin{array}{c}\mathrm{pBayer}\left[i\right]<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{lowval}\\ \left|\mathrm{pBayer}\right[0]-\mathrm{pBayer}[i\left]\right|<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{sameval}\end{array}\right.---\left(3\right)$

Wherein, pBayer [i] is the pixel value of right side with color pixel being positioned at same row with pixel to be detected, reg_hvdpc_lowval is the lower limit of the pixel value threshold value of setting, and reg_hvdpc_sameval is the absolute value threshold value of setting, the pixel value that pBayer [0] is pixel to be detected.

Or, for pixel to be detected in detection window and with pixel to be detected on the right side of same a line with color pixel i (0≤i≤2m, and i is even number) pixel value, if meet formula (4), the right side then representing pixel to be detected and go together with pixel to be detected is all very large with the brightness of color pixel, and the right side of going together with pixel to be detected is very little with the difference of the pixel value of color pixel and pixel to be detected:

$\left\{\begin{array}{c}\mathrm{pBayer}\left[i\right]>\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{highval}\\ \left|\mathrm{pBayer}\right[0]-\mathrm{pBayer}[i\left]\right|<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{sameval}\end{array}\right.---\left(4\right)$

Wherein, pBayer [i] is the pixel value of right side with color pixel being positioned at same row with pixel to be detected, reg_hvdpc_highval is the higher limit of the pixel value threshold value of setting, reg_hvdpc_sameval is the absolute value threshold value of setting, the pixel value that pBayer [0] is pixel to be detected.

When pixel to be detected and go together with pixel to be detected and the same color pixel be positioned on the right side of pixel to be detected all meets formula (3) or formula (4) time, then determine that this pixel to be detected is doubtful row bad point.

Doubtful row bad point determining unit 703, specifically for: for pixel to be detected in detection window and with pixel to be detected in the below of same row with color pixel i (0≤i≤2m, and i is even number) pixel value, if meet formula (5), then represent pixel to be detected and all very little with the brightness of color pixel with the downside of pixel same column to be detected, and very little with the difference of the pixel value of color pixel and pixel to be detected with the downside of pixel same column to be detected:

$\left\{\begin{array}{c}\mathrm{pBayer}[i*W]<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{lowval}\\ \left|\mathrm{pBayer}\right[0]-\mathrm{pBayer}[i*W\left]\right|<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{sameval}\end{array}\right.---\left(5\right)$

Wherein, pBayer [i*W] is the pixel value of below with color pixel being positioned at same row with pixel to be detected, reg_hvdpc_lowval is the lower limit of the pixel value threshold value of setting, reg_hvdpc_sameval is the absolute value threshold value of setting, the pixel value that pBayer [0] is pixel to be detected, W is the width for pixel place to be detected two field picture.

Or, for pixel to be detected in detection window and with pixel to be detected in the below of same row with color pixel i (0≤i≤2m, and i is even number) pixel value, if meet formula (6), then represent pixel to be detected and all very large with the brightness of color pixel with the downside of pixel same column to be detected, and very little with the difference of the pixel value of color pixel and pixel to be detected with the downside of pixel same column to be detected:

$\left\{\begin{array}{c}\mathrm{pBayer}[i*W]>\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{highval}\\ \left|\mathrm{pBayer}\right[0]-\mathrm{pBayer}[i*W\left]\right|<\mathrm{reg}\_\mathrm{hvdpc}\_\mathrm{sameval}\end{array}\right.---\left(6\right)$

Wherein, pBayer [i*W] is the pixel value of below with color pixel being positioned at same row with pixel to be detected, reg_hvdpc_highval is the higher limit of the pixel value threshold value of setting, reg_hvdpc_sameval is the absolute value threshold value of setting, the pixel value that pBayer [0] is pixel to be detected.

When pixel to be detected and with pixel same column to be detected and the same color pixel be positioned on the downside of pixel to be detected all meets formula (5) or formula (6) time, then determine that this pixel to be detected is doubtful row bad point.

Ranks bad point determining unit 704, specifically for: judge whether the first quantitative value is greater than the first amount threshold of setting, when determining that the first quantitative value is greater than the first amount threshold of setting, and determine that pixel to be detected is doubtful row bad point, then determine that this pixel to be detected is row bad point; And/or, when determining that the first quantitative value is greater than the first amount threshold of setting, and determine that pixel to be detected is doubtful row bad point, then determine that this pixel to be detected is row bad point.

When determining that the first quantitative value is not more than the first amount threshold of setting, then carry out conventional bad point detection.When determining that the first quantitative value is greater than the first amount threshold of setting, but determine that pixel to be detected is not doubtful row bad point, and be not doubtful row bad point, then carry out conventional bad point detection.

As shown in Figure 8, this device also comprises conventional dead pixel detecting unit 801, bad point correcting unit 802.Conventional dead pixel detecting unit 801, specifically for: add up the quantity that described pixel value difference is greater than the same color pixel of the second pixel value difference threshold value, as the second quantitative value, judge whether the second quantitative value is greater than the second amount threshold of setting; If the second quantitative value is greater than the second amount threshold of setting, then determine that pixel to be detected is conventional bad point, otherwise determine that pixel to be detected is normal pixel, detection trigger window moves.Wherein, the second pixel value difference threshold value is greater than the first pixel value difference threshold value under normal circumstances.

Bad point correcting unit 802, specifically for: after determining that pixel to be detected is bad point, determine the horizontal direction Grad of pixel to be detected, longitudinal direction Grad, left diagonal Grad, right diagonal Grad, choosing the direction that in above-mentioned 4 directions, direction gradient value is minimum is reference direction, as shown in Figure 9, it is the direction of 4 acquisition direction gradient values in a detection window, wherein black bars is pixel to be detected, hdir is Grad horizontal direction, ldir is the left diagonal of Grad, rdir is the right diagonal of Grad, vdir is gradient longitudinal direction, determine that pixel to be detected is after row bad point, determine the longitudinal direction Grad of pixel to be detected, left diagonal Grad, right diagonal Grad, choosing the direction that in above-mentioned 3 directions, direction gradient value is minimum is reference direction, determine that pixel to be detected is after row bad point, determine the horizontal direction Grad of pixel to be detected, left diagonal Grad, right diagonal Grad, choosing the direction that in above-mentioned 3 directions, direction gradient value is minimum is reference direction, determining that pixel to be detected is row bad point, is again after row bad point, determines the left diagonal Grad of pixel to be detected, right diagonal Grad, and choosing the direction that in above-mentioned 2 directions, direction gradient value is minimum is reference direction.Average in the reference direction determined with the pixel value of same color pixel of pixel to be detected at a distance of 2p pixel unit, using the mean value that obtains as the pixel value after pixel correction to be detected, wherein p is positive integer.Can average in reference direction in detection window, as the pixel value after pixel correction to be detected with two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit; Also the pixel value of whole pixels homochromy with pixel to be detected in reference direction in detection window can be averaged, as the pixel value after pixel correction to be detected.

To average in reference direction in detection window with two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit, be example as the pixel value after pixel correction to be detected, if the reference direction that pixel to be detected is determined is horizontal direction, then the pixel value after pixel correction to be detected is such as formula shown in (7):

dpc_pBayer＝(pBayer[2p]+pBayer[-2p])/2(7)

Wherein, dpc_pBayer is the pixel value after pixel correction to be detected, pBayer [2p] and pBayer [-2p] for pixel to be detected in a lateral direction with pixel to be detected at a distance of two pixel values with color pixel of 2p pixel unit.

If the reference direction that pixel to be detected is determined is longitudinal direction, then the pixel value after pixel correction to be detected is such as formula shown in (8):

dpc_pBayer＝(pBayer[2p*W]+pBayer[-2p*W])/2(8)

Wherein, dpc_pBayer is the pixel value after pixel correction to be detected, pBayer [2p*W] and pBayer [-2p*W] on pixel longitudinal direction to be detected with pixel to be detected at a distance of two pixel values with color pixel of 2p pixel unit, W is the width of pixel place to be detected two field picture.

If the reference direction that pixel to be detected is determined is left diagonal, then the pixel value after pixel correction to be detected is such as formula shown in (9):

dpc_pBayer＝(pBayer[2p*(W+1)]+pBayer[-2p*(W+1)])/2(9)

Wherein, dpc_pBayer is the pixel value after pixel correction to be detected, pBayer [2p* (W+1)] and pBayer [-2p* (W+1)] in the left diagonal of pixel to be detected with pixel to be detected at a distance of two pixel values with color pixel of 2p pixel unit, W is the width of pixel place to be detected two field picture.

If the reference direction that pixel to be detected is determined is right diagonal, then the pixel value after pixel correction to be detected is such as formula shown in (10):

dpc_pBayer＝(pBayer[2p*(W-1)]+pBayer[-2p*(W-1)])/2(10)

Wherein, dpc_pBayer is the pixel value after pixel correction to be detected, pBayer [2p* (W-1)] and pBayer [-2p* (W-1)] in the right diagonal of pixel to be detected with pixel to be detected at a distance of two pixel values with color pixel of 2p pixel unit, W is the width of pixel place to be detected two field picture.

After completing pixel correction to be detected, write in pixel to be detected by the pixel value after correcting, the pixel value making next detection window obtain pixel to be detected in current detection window is the pixel value after correcting, and detection trigger window moves.

Determine that the method for the direction gradient value of pixel to be detected is as follows: such as formula to determine shown in (11) in detection window with pixel to be detected in same a line and with two absolute values with the pixel value difference of color pixel of pixel to be detected at a distance of 2p pixel unit, using the horizontal direction Grad of the absolute value of this pixel value difference as pixel to be detected;

hdir＝|pBayer[2p]-pBayer[-2p]|(11)

Wherein, hdir is the horizontal direction Grad of pixel to be detected, and pBayer [2p] and pBayer [-2p] is and two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit.

Such as formula to determine shown in (12) in detection window with pixel to be detected at same row and with two absolute values with the pixel value difference of color pixel of picture to be detected at a distance of 2p pixel unit, using the longitudinal direction Grad of the absolute value of this pixel value difference as pixel to be detected;

vdir＝|pBayer[2p*W]-pBayer[-2p*W]|(12)

Wherein, vdir is the longitudinal direction Grad of pixel to be detected, and pBayer [2p*W] and pBayer [-2p*W] is and two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit.

Such as formula to determine shown in (13) on the left diagonal of detection window and with two absolute values with the pixel value difference of color pixel of pixel to be detected at a distance of 2p pixel unit, using the left diagonal Grad of the absolute value of this pixel value difference as pixel to be detected;

ldir＝|pBayer[2p*(W+1)]-pBayer[-2p*(W+1)]|(13)

Wherein, ldir is the longitudinal direction Grad of pixel to be detected, and pBayer [2p* (W+1)] and pBayer [-2p* (W+1)] is and two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit.

Shown in (14), to determine on the right diagonal of detection window and with two absolute values with the pixel value difference of color pixel of pixel to be detected at a distance of 2p pixel unit, using the right diagonal Grad of the absolute value of this pixel value difference as pixel to be detected;

rdir＝|pBayer[2p*(W-1)]-pBayer[-2p*(W-1)]|(14)

Wherein, rdir is the longitudinal direction Grad of pixel to be detected, and pBayer [2p* (W-1)] and pBayer [-2p* (W-1)] is and two pixel values with color pixel of pixel to be detected at a distance of 2p pixel unit.

Preferably, in above-described embodiment calculated direction Grad and correct after pixel value time, when p gets 1, be one preferably execution mode, but p be other integer values, and also can meet the embodiment of the present invention require value, also the present invention for protect scope in.

In the embodiment of the present invention, detection window is according to from left to right, order from top to bottom moves, when determining doubtful row bad point and doubtful row bad point, go together with pixel to be detected in detection window, and the pixel be positioned on the left of pixel to be detected, and with pixel same column to be detected, and the pixel be positioned on the upside of pixel to be detected is all the pixel after correcting, or be not the pixel of bad point, when therefore pixel to be detected being detected in the embodiment of the present invention, checking and pixel to be detected are gone together, and the pixel be positioned on the right side of pixel to be detected, and with pixel same column to be detected, and the pixel be positioned on the downside of pixel to be detected, when detection window moves according to other orders, when pixel to be detected is detected, determine whether pixel to be detected is dead pixel points of images according to the pixel do not detected.As shown in figures 10 a and 10b, carry out dead pixel points of images detection and correct the contrast schematic diagram of front and back for adopting the device in the embodiment of the present invention, Figure 10 a is the detected image schematic diagram before detecting and correcting, and Figure 10 b is the detected image schematic diagram after detecting and correcting.

Those skilled in the art should understand, embodiments of the invention can be provided as method, system or computer program.Therefore, the present invention can adopt the form of complete hardware embodiment, completely software implementation or the embodiment in conjunction with software and hardware aspect.And the present invention can adopt in one or more form wherein including the upper computer program implemented of computer-usable storage medium (including but not limited to magnetic disc store, CD-ROM, optical memory etc.) of computer usable program code.

The present invention describes with reference to according to the flow chart of the method for the embodiment of the present invention, equipment (system) and computer program and/or block diagram.Should understand can by the combination of the flow process in each flow process in computer program instructions realization flow figure and/or block diagram and/or square frame and flow chart and/or block diagram and/or square frame.These computer program instructions can being provided to the processor of all-purpose computer, special-purpose computer, Embedded Processor or other programmable data processing device to produce a machine, making the instruction performed by the processor of computer or other programmable data processing device produce device for realizing the function of specifying in flow chart flow process or multiple flow process and/or block diagram square frame or multiple square frame.

These computer program instructions also can be stored in can in the computer-readable memory that works in a specific way of vectoring computer or other programmable data processing device, the instruction making to be stored in this computer-readable memory produces the manufacture comprising command device, and this command device realizes the function of specifying in flow chart flow process or multiple flow process and/or block diagram square frame or multiple square frame.

These computer program instructions also can be loaded in computer or other programmable data processing device, make on computer or other programmable devices, to perform sequence of operations step to produce computer implemented process, thus the instruction performed on computer or other programmable devices is provided for the step realizing the function of specifying in flow chart flow process or multiple flow process and/or block diagram square frame or multiple square frame.

Although describe the preferred embodiments of the present invention, those skilled in the art once obtain the basic creative concept of cicada, then can make other change and amendment to these embodiments.So claims are intended to be interpreted as comprising preferred embodiment and falling into all changes and the amendment of the scope of the invention.

Obviously, those skilled in the art can carry out various change and modification to the present invention and not depart from the spirit and scope of the present invention.Like this, if these amendments of the present invention and modification belong within the scope of the claims in the present invention and equivalent technologies thereof, then the present invention is also intended to comprise these change and modification.

Claims (10)

1. a detection method for dead pixel points of images, is characterized in that, the method comprises:

For detected image, using the center of pixel to be detected as detection window, obtain the pixel value of described pixel to be detected and the same color pixel identical with described pixel color to be detected in detection window, and determine each pixel value difference with color pixel in described pixel to be detected and detection window respectively; Add up the quantity that described pixel value difference is greater than the same color pixel of the first pixel value difference threshold value, as the first quantitative value;

Determine described pixel to be detected in described detection window and with described pixel to be detected in same a line and the pixel value threshold range that all do not setting of the pixel value being positioned at the same color pixel on the right side of described pixel to be detected, and the absolute value of the pixel value difference of same color pixel on the right side of the described pixel to be detected determined and described pixel to be detected is all less than the absolute value threshold value of setting, then determine that described pixel to be detected is the doubtful row bad point of described detected image;

Determine described pixel to be detected in described detection window and with described pixel to be detected at same row and the pixel value threshold range that all do not setting of the pixel value being positioned at the same color pixel below described pixel to be detected, and the absolute value of the difference of the pixel value of same color pixel below the described pixel to be detected determined and described pixel to be detected is less than the absolute value threshold value of setting, then determine that described pixel to be detected is the doubtful row bad point of described detected image;

Determine that described first quantitative value is greater than the first amount threshold of setting, and determine that described pixel to be detected is the doubtful row bad point of described detected image, then determine that described pixel to be detected is the row bad point of described detected image; Determine that described first quantitative value is greater than the first amount threshold of setting, and determine that described pixel to be detected is the doubtful row bad point of described detected image, then determine that described pixel to be detected is the row bad point of described detected image.

2. the method for claim 1, is characterized in that, the method also comprises:

Determine that described first quantitative value is not more than the first amount threshold of setting; And/or, determine that described pixel to be detected is not doubtful row bad point, neither doubtful row bad point;

Trigger and conventional bad point detection is carried out to described pixel to be detected.

3. method as claimed in claim 2, is characterized in that, describedly carries out conventional bad point detection to pixel to be detected, specifically comprises:

Add up each pixel value difference with color pixel in described pixel to be detected and detection window and be greater than the quantity of the same color pixel of the second pixel value difference threshold value, as the second quantitative value, if described second quantitative value is greater than the second amount threshold of setting, then determine that described pixel to be detected is conventional bad point, otherwise determine that described pixel to be detected is normal pixel, detection trigger window moves.

4. the method as described in as arbitrary in claims 1 to 3, it is characterized in that, after determining that described pixel to be detected is row bad point, the longitudinal direction Grad of pixel to be detected described in described detection window, left diagonal Grad, right diagonal Grad are compared, determines that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, after determining that described pixel to be detected is row bad point, then the horizontal direction Grad of pixel to be detected described in described detection window, left diagonal Grad, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, determine that described pixel to be detected is row bad point, after being also row bad point, then the left diagonal Grad of pixel to be detected described in described detection window, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, determine that described pixel to be detected is not row bad point, neither after row bad point, then the horizontal direction Grad of pixel to be detected described in described detection window, longitudinal direction Grad, left diagonal Grad, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer.

5. method as claimed in claim 4, is characterized in that, determine that the method for direction gradient value comprises:

To determine in described detection window with pixel to be detected in same a line and with a pair pixel value with color pixel of pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the horizontal direction Grad of described absolute value as pixel to be detected;

To determine in described detection window with pixel to be detected at same row and with a pair pixel value with color pixel of pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the longitudinal direction Grad of described absolute value as pixel to be detected;

To determine in described detection window on left diagonal and with a pair pixel value with color pixel of described pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the left diagonal Grad of described absolute value as pixel to be detected;

To determine in described detection window on right diagonal and with a pair pixel value with color pixel of described pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the right diagonal Grad of described absolute value as pixel to be detected.

6. a checkout gear for dead pixel points of images, is characterized in that, this device comprises:

Pixel value statistic unit, for for detected image, using the center of pixel to be detected as detection window, obtain the pixel value of described pixel to be detected and the same color pixel identical with described pixel color to be detected in detection window, and determine each pixel value difference with color pixel in described pixel to be detected and detection window respectively; Add up the quantity that described pixel value difference is greater than the same color pixel of the first pixel value difference threshold value, as the first quantitative value;

Doubtful row bad point determining unit, for determining described pixel to be detected in described detection window and being positioned in same a line the pixel value threshold range that the homochromy pixel value on the right side of described pixel to be detected all do not setting with described pixel to be detected, and the absolute value of the pixel value difference of same color pixel on the right side of the described pixel to be detected determined and described pixel to be detected is all less than the absolute value threshold value of setting, then determine that described pixel to be detected is the doubtful row bad point of described detected image;

Doubtful row bad point determining unit, determine described pixel to be detected in described detection window and with described pixel to be detected at same row and the pixel value threshold range that all do not setting of the pixel value being positioned at the same color pixel below described pixel to be detected, and the absolute value of the difference of the pixel value of same color pixel below the described pixel to be detected determined and described pixel to be detected is less than the absolute value threshold value of setting, then determine that described pixel to be detected is the doubtful row bad point of described detected image;

Ranks bad point determining unit, for determining that described first quantitative value is greater than the first amount threshold of setting, and determines that described pixel to be detected is the doubtful row bad point of described detected image, then determines that described pixel to be detected is the row bad point of described detected image; Determine that described first quantitative value is greater than the first amount threshold of setting, and determine that described pixel to be detected is the doubtful row bad point of described detected image, then determine that described pixel to be detected is the row bad point of described detected image.

7. device as claimed in claim 6, is characterized in that, described ranks bad point determining unit, also for:

Determine that described first quantitative value is not more than the first amount threshold of setting; And/or, determine that described pixel to be detected is not doubtful row bad point, neither doubtful row bad point; Trigger and conventional bad point detection is carried out to described pixel to be detected.

8. device as claimed in claim 7, it is characterized in that, this device also comprises conventional dead pixel detecting unit, for:

Add up each pixel value difference with color pixel in described pixel to be detected and detection window and be greater than the quantity of the same color pixel of the second pixel value difference threshold value, as the second quantitative value, if described second quantitative value is greater than the second amount threshold of setting, then determine that described pixel to be detected is conventional bad point, otherwise determine that described pixel to be detected is normal pixel, detection trigger window moves.

9. the device as described in as arbitrary in claim 6 ~ 8, it is characterized in that, this device also comprises bad point correcting unit, specifically for:

After determining that described pixel to be detected is row bad point, then the longitudinal direction Grad of pixel to be detected described in described detection window, left diagonal Grad, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, after determining that described pixel to be detected is row bad point, then the horizontal direction Grad of pixel to be detected described in described detection window, left diagonal Grad, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, determine that described pixel to be detected is row bad point, after being also row bad point, then the left diagonal Grad of pixel to be detected described in described detection window, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer;

Or, determine that described pixel to be detected is not row bad point, neither after row bad point, then the horizontal direction Grad of pixel to be detected described in described detection window, longitudinal direction Grad, left diagonal Grad, right diagonal Grad are compared, determine that the minimum direction of Grad is reference direction; Average in the reference direction determined at a distance of a pair pixel value with color pixel of 2p pixel unit with described pixel to be detected, using the mean value determined as the pixel value after pixel correction to be detected, wherein p is positive integer.

10. device as claimed in claim 9, is characterized in that, described bad point correcting unit specifically for:

To determine in described detection window with pixel to be detected in same a line and with a pair pixel value with color pixel of pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the horizontal direction Grad of described absolute value as pixel to be detected;

To determine in described detection window with pixel to be detected at same row and with a pair pixel value with color pixel of pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the longitudinal direction Grad of described absolute value as pixel to be detected;

To determine in described detection window on left diagonal and with a pair pixel value with color pixel of described pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the left diagonal Grad of described absolute value as pixel to be detected;

To determine in described detection window on right diagonal and with a pair pixel value with color pixel of described pixel to be detected at a distance of 2p pixel unit, determine two pixel values are made difference take absolute value, using the right diagonal Grad of described absolute value as pixel to be detected.