CN109360189B - Method for detecting pixel defect point of image of uncooled infrared movement - Google Patents

Abstract

The invention discloses a method for detecting an image pixel defect point of an uncooled infrared movement, which comprises the following steps: acquiring N frames of core images through an image acquisition card, reading the acquired N frames of core images, and analyzing the gray value of each frame of core image to obtain the gray information of each frame of core image; averaging the gray values of the pixels corresponding to the N frames of the movement images to obtain an average image; and calculating a comparison area median difference MedDif [ x ] [ y ] of pixel coordinates (x, y) in the average image and a judgment threshold BP of the current state of the movement, judging the pixel as a bright defect point if MedDif [ x ] [ y ] is larger than BP, judging the pixel as a dark defect point if MedDif [ x ] [ y ] is smaller than BP, and otherwise judging the pixel as not a pixel defect point. The method can objectively, accurately and quickly identify the defective points on the image, and can accurately judge the pixel size occupied by each defective point and the position coordinates of the defective points.

Description

Method for detecting pixel defect point of image of uncooled infrared movement

Technical Field

The invention relates to the technical field of non-refrigeration infrared movement image pixel defect points (bright or dark pixel points which are different from numerous surrounding pixel points and are expressed in a non-refrigeration infrared movement image, and are referred to as defect points for short), in particular to a method for detecting the non-refrigeration infrared movement image pixel defect points based on an image acquisition card.

Background

The image quality of the movement is an important index of interest to the user, and the number and size of the image defect points are one of important indexes for evaluating the image quality.

The original method for detecting the defect points of the infrared movement image is characterized in that human eyes recognize the defect points, each parameter of the movement is adjusted to be moderate, a uniform background is provided for the movement, the number of the defect points on the image and the pixel size occupied by the defect points are recognized by the human eyes, and the approximate area where the defect points are located is recorded. It is feasible for human eyes to recognize the number of defect points, but it is difficult to accurately recognize the pixel size and the position coordinates occupied by each defect point, and only an approximate judgment can be obtained by experience. The human eye recognition has certain subjectivity, and the detection accuracy is not enough, so that the detection efficiency is low.

Disclosure of Invention

The invention provides a method for detecting pixel defect points of an image of an uncooled infrared movement based on an image acquisition card, aiming at the current situation of detecting the image defect points of the uncooled infrared movement, which can objectively, accurately and quickly identify the defect points on the image and accurately judge the pixel size occupied by each defect point and the position coordinates of the defect points.

The invention solves the technical problems through the following technical scheme:

the invention provides a method for detecting an image pixel defect point of an uncooled infrared movement, which is characterized by comprising the following steps of:

s1, acquiring N frames of core images through an image acquisition card, reading the acquired N frames of core images, and analyzing the gray value of the N frames of core images to obtain the gray information of each frame of core image, wherein N is a positive integer;

s2, averaging and calculating the corresponding pixel gray values of the N frames of core images to obtain an average image;

s3, judging whether the core image has a defect point by analyzing the average image, calculating the median difference MedDif [ x ] [ y ] in the contrast area of the pixel coordinate (x, y) in the average image and the judgment threshold BP of the current state of the core, if MedDif [ x ] [ y ] is larger than BP, the pixel is judged as a bright defect point, if MedDif [ x ] [ y ] is smaller than BP, the pixel is judged as a dark defect point, otherwise the pixel is judged as not a pixel defect point.

Preferably, in step S3, when determining whether the pixel with coordinates (x, y) belongs to the defect point, the gray level of a region of 15 × 15 pixels around the pixel is determined, and the region is referred to as a contrast region of the pixel with coordinates (x, y);

using formulas

Calculating Gray value of coordinate (x, y) pixel in average image, wherein Gray [ n ]][x][y]Is the gray value of the pixel with the coordinate of (x, y) in the image of the nth frame, VGray [ x [ ]][y]The gray value of a pixel with coordinates (x, y) in the average image is represented, and N is less than or equal to N;

sorting all pixel gray values in a contrast area of pixel coordinates (x, y) in an average image from small to large, and calling the pixel gray value arranged at the middle as a median Med [ x ] [ y ] of the contrast area;

and calculating the value difference in the contrast area by using the formula MedDif [ x ] [ y ] (VGray [ x ] [ y ] -Med [ x ] [ y ].

Preferably, in step S3, the average of all gray values of pixels in the contrast area of the pixel coordinate (x, y) in the average image is called the average Aver [ x ] [ y ] of the contrast area, the standard deviation of all gray values of pixels in the contrast area of the pixel coordinate (x, y) in the average image is defined as the uniformity Un [ x ] [ y ] of the contrast area, the average of uniformity of the contrast area corresponding to all pixels in the average image is called the uniformity Un _ Aver:

wherein: xmax represents the number of horizontal pixels of the image, and ymax represents the number of vertical pixels of the image;

the judgment threshold is calculated by using the formula BP 8.196 Un _ AVER.

On the basis of the common knowledge in the field, the above preferred conditions can be combined randomly to obtain the preferred embodiments of the invention.

The positive progress effects of the invention are as follows:

the invention relates to a detection method of an image defect point of an uncooled infrared movement, which is researched by combining methods such as an image video acquisition card technology, a computer programming technology, a data statistical analysis method, mathematical modeling and the like and can objectively, quickly, accurately and automatically detect the image defect point of the movement. The method solves the problems of accurate positioning of the position coordinates of the defect points, definite division of the areas where the defect points are located, statistics of the pixel sizes occupied by the continuous defect points and the like.

Drawings

Fig. 1 is a flowchart of a method for detecting a pixel defect point in an image of an uncooled infrared cassette mechanism according to a preferred embodiment of the invention.

FIG. 2 is a diagram of the contrast area of a pixel with coordinates (x, y) according to the preferred embodiment of the present invention.

Detailed Description

Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention. On the contrary, the embodiments of the invention include all changes, modifications and equivalents coming within the spirit and terms of the claims appended hereto.

As shown in fig. 1, the present embodiment provides a method for detecting a pixel defect point of an image of an uncooled infrared movement, which includes the following steps:

step 101, acquiring N frames of core images through an image acquisition card, reading the acquired N frames of core images, and analyzing the gray value of the N frames of core images to obtain the gray information of each frame of core image, wherein N is a positive integer;

102, averaging and calculating corresponding pixel gray values of the N frames of the core images to obtain an average image;

and 103, judging whether the core image has a defect point or not by analyzing the average image, calculating a comparison area median difference MedDif [ x ] [ y ] of pixel coordinates (x, y) in the average image and a judgment threshold BP of the current state of the core, judging the pixel as a bright defect point if MedDif [ x ] [ y ] is larger than BP, judging the pixel as a dark defect point if MedDif [ x ] [ y ] is smaller than BP, and otherwise judging the pixel as not a pixel defect point.

The following describes specifically determining whether the movement image has a defect point by analyzing the average image information.

When determining whether the pixel with the coordinate (x, y) belongs to the defect point, the gray values of 15 × 15 pixel regions around the pixel point should be combined to determine, as shown in fig. 2, the region is referred to as a contrast region with the coordinate (x, y) as a contrast region, which is referred to as a contrast region for short.

First, the derivation calculation method of the value difference (MedDif [ x ] [ y ]) in the contrast area

Defining Gray [ N ] [ x ] [ y ] as the Gray value of the pixel with the coordinate (x, y) in the nth frame image, VGray [ x ] [ y ] as the Gray value of the pixel with the coordinate (x, y) in the average image, and setting the total frame number of the acquired video as N, then:

equation 1:

in the average image, all pixel gray values in the contrast area of the pixel coordinate (x, y) in the average image are sorted from small to large, the gray value of the pixel arranged at the middle is called the contrast area median, which represents the general level of the gray level of all pixels in the contrast area, and is expressed by Med [ x ] [ y ], the difference value between VGray [ x ] [ y ] and Med [ x ] [ y ] is called the contrast area median difference, which represents the difference level of the gray value of the pixel and the gray level of the pixel around the pixel, and is expressed by Med Dif [ x ] [ y ]:

equation 2:

Med[x][y]＝Median{VGray[x+n][y+m]|n,m＝-7,-6,-5,-4,-3,-2,-1,0,1,2,3,4,5,6,7}

equation 3: MedDif [ x ] [ y ] (VGray [ x ] [ y ] -Med [ x ] [ y ]

If the pixel point with the coordinate (x, y) is not a defect point, the absolute value of the value difference MedDif [ x ] [ y ] in the contrast area is very small and approaches to 0, otherwise, when MedDif [ x ] [ y ] is larger than BP, the point belongs to a bright defect point, and when MedDif [ x ] [ y ] is smaller than-BP, the point belongs to a dark defect point, and BP is called a judgment threshold.

Determination method of judgment threshold (BP)

In the average image, the average value of all pixel grays in the contrast area of the pixel coordinate (x, y) in the average image is called the contrast area average value and is expressed by Aver [ x ] [ y ], and the standard deviation of all pixel grays in the contrast area of the pixel coordinate (x, y) in the average image is defined as the contrast area uniformity, which represents the degree of dispersion between all pixel grays in the contrast area and is expressed by Un [ x ] [ y ]:

equation 4:

equation 5:

in the whole image, the average value of the uniformity of the contrast area corresponding to all pixels is called the average uniformity of the image, which is called the average uniformity for short, and it represents the uniformity degree between the gray levels of the pixels of the whole image, and is expressed by Un _ AVER:

equation 6:

wherein: xmax represents the number of horizontal pixels of the image; ymax represents the number of vertical pixels of the image.

In order to determine the light-dark threshold BP, the following tests were performed using cartridges known to have defective spots, as shown in tables 1 and 2 below for cartridges of types 384 × 288 and 640 × 512, respectively. The "state" in the table refers to the state of the same movement under different contrasts, and the "defect point median difference detailed graph" is an enlarged view of the defect point, and the number in the graph is the median difference of the corresponding coordinate point.

The following rules are found from the data in tables 1 and 2: firstly, the larger the gray difference between the defective pixel point and the normal pixel in the contrast area is, the larger the absolute value of the median difference corresponding to the pixel is, and otherwise, the smaller the absolute value of the median difference is; secondly, for the same core, when the average uniformity Un _ AVER is larger, the median difference of the same defective pixel point on the core image is also larger, which indicates that the median difference between Un _ AVER and the defective pixel point has positive correlation; third, one defective dot is composed of one or more defective pixels, the center gray difference of the defective dot is the largest, and the gray difference of the pixels from the center to the outside becomes smaller step by step.

TABLE 1

TABLE 1

As can be seen from tables 1 and 2 above, a defect dot is generally composed of two or more pixel defect dots. The minimum median difference in a defect point is defined as MF _ MIN, and further tests were performed with known cartridges having defect points in order to prove that the decision threshold BP is positively correlated with the average uniformity Un _ AVER, the test results being shown in tables 3 and 4.

The coefficient MIN _ AK in tables 3 and 4 is the ratio of MF _ MIN to Un _ AVER, i.e.:

equation 7: MIN _ AK ═ MF _ MIN/Un _ AVER;

taking the movement numbered MDIC052-154 as an example, in 3 states, the BP when the gain is small should be less than 9, the BP when the gain is medium should be less than 13, the BP when the gain is large should be less than 16, the software detection result can accord with the human eye detection, and the BP is increased along with the increase of the gain, and the BP is not a constant value. From tables 3 and 4, it can be seen that as the gain of the core increases, the average uniformity Un _ AVER increases, which indicates that there is a certain relationship between the judgment threshold BP and Un _ AVER.

Assume equation 8: BP ═ Un _ AVER × AK

Wherein AK is a positive correlation coefficient between the judgment threshold and the average uniformity.

Equation 9 is derived from equation 7: MF _ MIN ═ Un _ AVER ═ MIN _ AK

To avoid missing detection of a defect point, BP must satisfy the relationship BP < MF _ MIN and can be derived from equations 8 and 9: equation 10: AK < MIN _ AK

TABLE 3

TABLE 4

Meanwhile, a great deal of experiments are also performed by using a known movement without defect points, the experimental conditions are detailed in tables 5 and 6, MF _ MAX in the tables represents the maximum value of the median difference of all pixel contrast areas in the average image of the movement, and the coefficient MAX _ AK is the ratio of MF _ MAX to Un _ AVER, that is:

equation 11: MF _ MAX ═ MAX { MedDif [ x ] [ y ] | x ═ 1:2: xmax, y ═ 1:2: ymax }

Equation 12: MAX _ AK ═ MF _ MAX/Un _ AVER

It can be known from the experimental data that the erroneous judgment of the defect point does not occur only when BP satisfies the relationship BP > MF _ MAX. Therefore, it can be derived from equations 8 and 12:

equation 13: AK > MAX _ AK

TABLE 5

TABLE 6

From the experimental data in table 3, table 4, table 5 and table 6, the MAX _ AK is 7.652, and the MIN _ AK is 8.739. From equation 10 and equation 13, MAX _ AK < MIN _ AK, i.e.: 7.652< AK <8.739, the assumption of equation 8 holds because 7.652< AK <8.739 is a true proposition.

According to a lot of experiments, it is reasonable to take the intermediate value between the maximum MAX _ AK and the minimum MIN _ AK for AK, i.e. AK is 8.196, and BP is 8.196 Un _ AVER according to formula 8.

According to the method described above, the value of AK can be determined by the same method for other types or models of machine cores, and then the value of the judgment threshold BP can be obtained.

According to the algorithm in the technical scheme, the objective, accurate and quick detection of the image defect point of the uncooled infrared movement is realized by combining the video acquisition card technology and programming through computer software.

Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.

Claims (1)

1. A method for detecting pixel defect points of an image of an uncooled infrared movement is characterized by comprising the following steps:

s1, acquiring N frames of core images through an image acquisition card, reading the acquired N frames of core images, and analyzing the gray value of the N frames of core images to obtain the gray information of each frame of core image, wherein N is a positive integer;

s2, averaging and calculating the corresponding pixel gray values of the N frames of core images to obtain an average image;

s3, judging whether the core image has a defect point or not by analyzing the average image, calculating the median difference MedDif [ x ] [ y ] in the contrast area of the pixel coordinate (x, y) in the average image and the judgment threshold BP of the current state of the core, if MedDif [ x ] [ y ] is larger than BP, the pixel is judged as a bright defect point, if MedDif [ x ] [ y ] is smaller than BP, the pixel is judged as a dark defect point, otherwise the pixel is judged as not a pixel defect point;

in step S3, when determining whether the pixel with coordinates (x, y) belongs to the defect point, the gray level of the 15 × 15 pixel regions around the pixel is determined, and the regions are called contrast regions of the pixel with coordinates (x, y);

using the formula VGray x][y]=

The Gray value of a pixel with coordinates (x, y) in the average image is calculated by/N, wherein Gray [ N ]][x][y]Is the gray value of the pixel with the coordinate of (x, y) in the image of the nth frame, VGray [ x [ ]][y]The gray value of a pixel with coordinates (x, y) in the average image is represented, and N is less than or equal to N;

sorting all pixel gray values in a contrast area of pixel coordinates (x, y) in an average image from small to large, and calling the pixel gray value arranged at the middle as a median Med [ x ] [ y ] of the contrast area;

calculating the value difference in the contrast area by using a formula MedDif [ x ] [ y ] = VGray [ x ] [ y ] -Med [ x ] [ y ];

in step S3, the average of all the gray values of the pixels in the contrast area of the pixel coordinate (x, y) in the average image is called the average Aver [ x ] [ y ] of the contrast area, the standard deviation of the gray values of all the pixels in the contrast area of the pixel coordinate (x, y) in the average image is defined as the uniformity Un [ x ] [ y ] of the contrast area, the average of the uniformity of the contrast area corresponding to all the pixels in the average image is called the uniformity Un _ Aver:

wherein: xmax represents the number of horizontal pixels of the image, and ymax represents the number of vertical pixels of the image;

the judgment threshold is calculated using the formula BP =8.196 × Un _ AVER.

Images