CN110717887A - Method for detecting bad elements of line-row detector - Google Patents

Abstract

The invention relates to a method for detecting a bad element of a line detector, belonging to the technical field of image processing. The technology combines a time domain method and a window method to realize the bad element detection of the linear infrared detector. The method is simple in principle, can effectively detect the bad elements in the linear infrared detector, and reduces the false detection rate.

Description

Method for detecting bad elements of line-row detector

Technical Field

The invention relates to a method for detecting a bad element of a line detector, belonging to the technical field of image processing.

Background

In an infrared line type imaging system, due to the problems of device production processes, an infrared line detector often has blind pixels and flash bad pixels, so that bright (dark) stripes (caused by the blind pixels) or alternatively bright and dark stripes (caused by the flash bad pixels) appear in an image along a scanning direction when the system is scanned and imaged, and the imaging quality and the subsequent small target detection effect are seriously affected. Especially, the flicker bad elements cause the generated stripes with alternating brightness and darkness, and the detection of the small target is greatly influenced. For the detection of bad elements of the linear detector, a time domain method and a window detection method are representative methods. The time domain method has higher detection precision, but cannot be realized in real time; the existing window method mainly detects blind pixels and cannot effectively detect scintillation bad pixels. Therefore, a bad element detection method combining a time domain method and a window method is provided to improve the imaging quality of the line array detector.

Disclosure of Invention

Technical problem to be solved

The invention provides a method for detecting a defective element of a linear array detector, aiming at solving the problem that a blind element detection algorithm in the conventional linear array detector can only detect a blind element and can not effectively detect a flicker defective element.

Technical scheme

A method for detecting a bad element of a linear array detector is characterized by comprising the following steps:

step 1: acquiring infrared image data and putting the infrared image data into a frame memory;

step 2: calculating the average value of each row of image data by taking P outputs as a group and storing the average value in an array R _ AVG [ M x N ]; wherein M is the number of detector pixels, and N is an integer of a quotient obtained by dividing the number of a line of image data by P;

and step 3: performing tail-biting average filtering on a first row R _ AVG [ M × 1] of an average array R _ AVG [ M × N ]: taking the window size as W1, carrying out tail-cutting average filtering on an average value array R _ AVG [ M1 ] in the sliding window, and storing a filtering result in an array R _ Mean [ M ];

and 4, step 4: calculating the difference value between the Mean value array R _ AVG [ M x 1] and the truncated Mean value filtering result R _ Mean [ M ], and storing the difference value in the array ERR _ AVG [ M ];

and 5: and judging the array ERR _ AVG [ M ] and a set threshold AVG _ TH: if the absolute value of the ith data in the array ERR _ AVG [ M ] is greater than the threshold value AVG _ TH, marking the ith pixel as a blind pixel;

step 6: calculating the standard deviation of the mean value array R _ AVG [ M × N ] according to rows, and storing the standard deviation in an array R _ ST [ M ];

and 7: median filtering the array of standard deviations R _ ST [ M ]: taking the window size as W x 1, carrying out Median filtering on a standard deviation array R _ ST [ M ] in the sliding window, and storing a filtering result in an array R _ Median [ M ];

and 8: calculating the difference value between the standard difference array R _ ST [ M ] and the Median filtering result R _ media [ M ] and storing the difference value in the array ERR [ M ];

and step 9: calculating an adaptive threshold: calculating the standard deviation sigma of the array ERR [ M ], and multiplying the standard deviation sigma by a set weight K to obtain a detection threshold Th of the flicker bad element;

step 10: and judging the array ERR [ M ] by using a threshold Th, if the absolute value of the ith data in the array ERR [ M ] is greater than the threshold Th, marking the ith pixel as a flicker bad element, otherwise, considering that the ith pixel is not the flicker bad element.

The truncated mean filtering method described in step 3 is as follows:

a) recording W mean values in the sliding window as an array AVG _ WD [ W ];

b) judging whether the numerical value of the window center is an extreme value, if not, calculating the difference value between the rest numerical values in the window and the central numerical value, and if the absolute value of the difference value is smaller than a threshold value W _ TH, putting the numerical value into an array AVG _ TEMP; if the numerical value of the center of the window is an extreme value, taking the number on the left side of the central numerical value as a reference, and putting the absolute value of the difference between the rest numerical values and the reference numerical value, which is smaller than the threshold value W _ TH, into the array AVG _ TEMP;

c) and calculating the average value of the array AVG _ TEMP and outputting the average value.

Advantageous effects

The method for detecting the line detector bad elements provided by the invention combines a time domain method and a window method, not only can effectively detect the blind elements, but also can effectively detect the scintillation bad elements, thereby greatly improving the blind element detection efficiency of the line detector and further improving the imaging effect of the line detector. The method is simple in principle, can effectively detect the bad elements in the linear infrared detector, and reduces the false detection rate.

Drawings

FIG. 1 is a flowchart of the procedure of the present invention

Detailed Description

The invention will now be further described with reference to the following examples and drawings:

the specific implementation method comprises the following steps:

(1) after the system operates, acquiring infrared image data and putting the infrared image data into a frame memory;

(2) for each line of image data, the average of the P outputs is calculated as a group and stored in an array R _ AVG [ M × N ]. Wherein M is the number of detector pixels, and N is an integer of a quotient obtained by dividing the number of a line of image data by P;

(3) and performing truncation average filtering on the first row R _ AVG [ M × 1] of the average array R _ AVG [ M × N ]. And taking the window size as W1, carrying out truncated Mean filtering on the Mean value array R _ AVG [ M1 ] in the sliding window, and storing the filtering result in the array R _ Mean [ M ].

(4) Calculating the difference value between the Mean value array R _ AVG [ M x 1] and the truncated Mean value filtering result R _ Mean [ M ], and storing the difference value in the array ERR _ AVG [ M ];

(5) setting a threshold value AVG _ TH, judging the array ERR _ AVG [ M ] by the threshold value AVG _ TH, and if the absolute value of the ith data in the array ERR _ AVG [ M ] is larger than the threshold value AVG _ TH, marking the ith pixel as a blind pixel;

(6) calculating the standard deviation of the mean value array R _ AVG [ M × N ] according to rows, and storing the standard deviation in an array R _ ST [ M ];

(7) the standard deviation array R _ ST [ M ] is median filtered. And taking the window size as W x 1, performing Median filtering on the standard deviation array R _ ST [ M ] in the sliding window, and storing the filtering result in the array R _ Median [ M ].

(8) Calculating the difference value between the standard difference array R _ ST [ M ] and the Median filtering result R _ media [ M ] and storing the difference value in the array ERR [ M ];

(9) an adaptive threshold is calculated. Calculating the standard deviation sigma of the array ERR [ M ], and multiplying the standard deviation sigma by a set weight K to obtain a detection threshold Th of the flicker bad element;

(10) and judging the array ERR [ M ] by using a threshold Th, if the absolute value of the ith data in the array ERR [ M ] is greater than the threshold Th, marking the ith pixel as a flicker bad element, otherwise, considering that the ith pixel is not the flicker bad element.

Example 1:

for a 480-pixel linear infrared imaging system, after the infrared system is powered on and operated, an infrared image with 480 × 640 pixels is obtained, namely, an infrared image with 480 rows and 640 columns. The image is averaged by row for a group of 16 data and stored in the array R _ AVG [480 x 40] (step 2). And (3) performing truncation Mean filtering on the first row R _ AVG [480 × 1] of the R _ AVG [480 × 40] by using a window 7 × 1 to obtain an array R _ Mean [480], wherein the threshold value W _ TH in the truncation Mean filtering is 10 (step 3). And calculating the difference between the array R _ AVG [480 x 1] and the array R _ Mean [480] to obtain a difference array ERR _ AVG [480] (step 4). Setting a threshold value AVG _ TH to be 16, judging the logarithm group ERR _ AVG [480] by using the threshold value AVG _ TH, and if the absolute value of the ith data in the logarithm group ERR _ AVG [480] is larger than the threshold value AVG _ TH, marking the ith pixel as a blind pixel (step 5). The standard deviation of the mean array R _ AVG [480 × 40] is calculated row by row and stored in the array R _ ST [480] (step 6). The standard deviation array R _ ST [480] is Median filtered with a window 9 x 1 to yield the array R _ Median [480] (step 7). The difference between the array of standard differences R _ ST [480] and the Median filter result R _ Median [480] is calculated and stored in the array ERR [480] (step 8). The standard deviation sigma of the array ERR [480] is calculated and then the adaptive threshold is found: th ═ K × sigma, where K ═ 3 (step 9). And judging the array ERR [480] by using a threshold Th, if the absolute value of the ith data in the array ERR [480] is greater than the threshold Th, marking the ith pixel as a flicker bad element, otherwise, considering the ith pixel as not a flicker bad element (step 10).

Claims (2)

1. A method for detecting a bad element of a linear array detector is characterized by comprising the following steps:

step 1: acquiring infrared image data and putting the infrared image data into a frame memory;

step 2: calculating the average value of each row of image data by taking P outputs as a group and storing the average value in an array R _ AVG [ M x N ]; wherein M is the number of detector pixels, and N is an integer of a quotient obtained by dividing the number of a line of image data by P;

and step 3: performing tail-biting average filtering on a first row R _ AVG [ M × 1] of an average array R _ AVG [ M × N ]: taking the window size as W1, carrying out tail-cutting average filtering on an average value array R _ AVG [ M1 ] in the sliding window, and storing a filtering result in an array R _ Mean [ M ];

and 4, step 4: calculating the difference value between the Mean value array R _ AVG [ M x 1] and the truncated Mean value filtering result R _ Mean [ M ], and storing the difference value in the array ERR _ AVG [ M ];

and 5: and judging the array ERR _ AVG [ M ] and a set threshold AVG _ TH: if the absolute value of the ith data in the array ERR _ AVG [ M ] is greater than the threshold value AVG _ TH, marking the ith pixel as a blind pixel;

step 6: calculating the standard deviation of the mean value array R _ AVG [ M × N ] according to rows, and storing the standard deviation in an array R _ ST [ M ];

and 7: median filtering the array of standard deviations R _ ST [ M ]: taking the window size as W x 1, carrying out Median filtering on a standard deviation array R _ ST [ M ] in the sliding window, and storing a filtering result in an array R _ Median [ M ];

and 8: calculating the difference value between the standard difference array R _ ST [ M ] and the Median filtering result R _ media [ M ] and storing the difference value in the array ERR [ M ];

and step 9: calculating an adaptive threshold: calculating the standard deviation sigma of the array ERR [ M ], and multiplying the standard deviation sigma by a set weight K to obtain a detection threshold Th of the flicker bad element;

step 10: and judging the array ERR [ M ] by using a threshold Th, if the absolute value of the ith data in the array ERR [ M ] is greater than the threshold Th, marking the ith pixel as a flicker bad element, otherwise, considering that the ith pixel is not the flicker bad element.

2. The method for detecting the bad elements of the line detector according to claim 1, wherein the truncated mean filtering method described in step 3 is as follows:

a) recording W mean values in the sliding window as an array AVG _ WD [ W ];

b) judging whether the numerical value of the window center is an extreme value, if not, calculating the difference value between the rest numerical values in the window and the central numerical value, and if the absolute value of the difference value is smaller than a threshold value W _ TH, putting the numerical value into an array AVG _ TEMP; if the numerical value of the center of the window is an extreme value, taking the number on the left side of the central numerical value as a reference, and putting the absolute value of the difference between the rest numerical values and the reference numerical value, which is smaller than the threshold value W _ TH, into the array AVG _ TEMP;

c) and calculating the average value of the array AVG _ TEMP and outputting the average value.

Images