CN-115797382-B - Frame difference local contrast based infrared weak and small moving object edge detection method

CN115797382BCN 115797382 BCN115797382 BCN 115797382BCN-115797382-B

Abstract

The invention relates to an infrared weak and small moving object edge detection method based on frame difference local contrast, which comprises the steps of determining the difference of the radiation characteristics among the main false alarm source high-brightness cloud layer edge, blind flash noise and an object, and enhancing the edge of the target by using a three-frame time domain difference and improved local contrast computing method, and removing false alarm sources such as a flashing element, a highlight background, a background edge and the like. And then obtaining a target edge detection result by threshold segmentation according to the generated saliency map, and finally returning to the original map to obtain a final accurate target position. The invention provides an infrared weak and small moving object edge detection method based on frame difference local contrast, which is based on the fact that the detection of a weak infrared object under a complex dynamic background in actual engineering is difficult to meet the requirements of detection rate and false alarm rate. Experimental results show that the algorithm has higher stability, effectively reduces the number of false alarms and improves the target detectability while ensuring the detection efficiency.

Inventors

LIU ZIJIAN
DING PENGYUAN
JIANG SHITONG
Gu Enchen
ZHANG YIN
YAN JUNHUA
ZHU DEYAN

Assignees

南京航空航天大学

Dates

Publication Date: 20260505
Application Date: 20221130

Claims (6)

1. An infrared weak and small moving object edge detection method based on frame difference local contrast is characterized by comprising the following steps: step one, obtaining a target image, and carrying out contrast enhancement on the target by adopting a difference method in three continuous frames of images to obtain a first difference feature map; step two, utilizing an improved local contrast computing method to inhibit background clutter in the image and strengthen the edge of a target, wherein the specific process of the step two is as follows: Using For the first differential feature map in step one Performing sequencing filtering of the whole graph, taking a gray level intermediate value as gray level estimation of a central target window, constructing an intermediate window to isolate a target background, performing local contrast calculation by utilizing the difference between the infrared weak small target edge area and the adjacent background area, and inhibiting background clutter while enhancing the target; Step three, target energy obtained by calculation through a central window is adopted to strengthen the target edge; the specific process of the third step is as follows: Calculation of The difference value between the maximum gray level value and the minimum gray level value in the sequencing filter, the noise and abnormal pixels on the time domain and the space domain can be removed by utilizing the edge dispersion characteristic of the target, the background edge is restrained, and the edge of the target is reinforced; And fourthly, removing unstable response points of the image by adopting self-adaptive threshold segmentation to obtain a detection result of the target diffuse edge, and returning the detection result of the target edge to the differential feature map for target positioning to obtain an accurate detection result.
2. The method for detecting the edge of the infrared small moving object based on the frame difference local contrast according to claim 1, wherein the specific process of the first step is as follows: Acquiring a target image, and calculating the difference value between a second frame and a third frame in three continuous frame images And the difference between the second frame and the first frame And re-deriving a first differential feature map for detection of a second frame image The contrast between the object and the background is enhanced using time domain noise instead of spatial domain noise.
3. The method for detecting the edge of the infrared small moving object based on the frame difference local contrast according to claim 2, wherein the first differential feature map Obtained by the following formula: 。
4. the method for detecting the edge of the infrared small moving object based on the frame difference local contrast according to claim 2, wherein in the first step, the background and the noise are suppressed by using threshold segmentation, and the threshold segmentation is specifically as follows: , Wherein, the Is a threshold value.
5. The method for detecting the edge of the infrared small moving object based on the frame difference local contrast according to claim 1, wherein the second step comprises the following steps: Step 2.1, decomposing the image into four pixels, namely an upper left, an upper right, a lower left and a lower right of the image, sliding the pixels from left to right and from top to bottom on the image by using a sliding window, dividing each window into three parts, wherein the central position is a region where a target edge possibly appears, the middle layer is a target protection layer, and the outermost layer is a background layer, dividing the background layer into 8 parts, and respectively calculating average pixel values of 8 neighborhoods as background estimation, namely ; Wherein, the Representing the number of pixels contained in the sub-window, Representing the effective value of the jth pixel within the ith sub-window; Step 2.2 when the edge of the object appears in the center position, at least 3 pixels respond by Calculating the pixel gray sum of the gray second and third gray third As a gray scale estimation for the target edge, and the pixel gray scale of the limit gray scale third cannot be 0: ; In the formula, The gray value of the picture element representing the second gray level, A pixel gray value representing a third gray; step 2.3, calculating the difference between the target edge sub-window and the background: - ; the adoption of the MPCM algorithm can well inhibit continuous background, and a second characteristic diagram C is obtained through calculation: ; Wherein, the Representing the position coordinates of the upper left picture element of the target area, Representing the contrast between the target edge sub-window and the background sub-window i at coordinates (m, n), Representing the contrast between the target edge sub-window and the background sub-window i +4 at coordinates (m, n).
6. The method for detecting the edge of the infrared small moving object based on the frame difference local contrast according to claim 2, wherein the third step comprises the following steps of: If the central window falls in the central position of the target or the background edge with a large area, the gray scale difference among the four pixels is not large; if the central window falls on the isolated noise point and the noise quantity is 1 or 2, the window is restrained in the second step; if the central window falls on the edge of the target, the gray value of the main pixel is the largest, the gray value of the outermost dispersed pixel is the smallest and is close to the background gray value; weighting the second feature map C by using the diffuse effect of the target edge: ; In the formula, Representing the position coordinates of the upper left picture element of the target area, Is a picture element The weighted characteristic values are specifically defined as follows: ; In the formula, And Respectively at pixel positions m, n A gray maximum value and a gray minimum value in the sliding window; Obtaining target gray scale estimation at pixel positions m and n in the second step; is a background sliding window mean value taking m and n as centers; Is a protection parameter.

Description

Frame difference local contrast based infrared weak and small moving object edge detection method Technical Field The invention relates to the technical field of infrared small target detection, in particular to an infrared weak small moving target edge detection method based on frame difference local contrast. Background The development of infrared technology makes it widely used in various fields such as early warning, monitoring, terminal guidance, etc. And infrared image target detection is one of the most widely applied and most urgent problems to be solved. Due to the limitation of infrared spectrum, compared with a visible light image, the infrared image has lower resolution and fewer available characteristics, the target is extremely easy to submerge in background clutter, and the detection is very difficult. Meanwhile, an infrared imaging system applied to a space base is influenced by factors such as high-rise cloud rolling, platform image shifting and the like, the background change is severe, and target detection is more difficult. Therefore, the detection of infrared dim targets in a dynamic complex background has been a hotspot and difficulty of research. At present, research on infrared image target detection is largely divided into two types, namely a single-frame-based infrared weak target detection method and a multi-frame-based infrared weak target detection method. Due to the dynamic change of the background, the image shift of the field of view and the uncertainty of the target movement, the infrared dim target detection technology based on the multi-frame method is limited in performance. Therefore, the current research mainly aims at single-frame infrared weak and small target detection, and can be roughly divided into a detection method based on filtering, a detection method based on a transform domain, a detection method based on matrix sparse low-rank decomposition and a detection method based on local contrast. However, at the same time, the single-frame target detection method requires that the target has local significance, so that the detection method is difficult to detect the weak target submerged in the complex background. Therefore, how to design a method for detecting a weak and small target under a complex dynamic background with high accuracy, low false detection and few false alarms is a problem to be solved urgently. Disclosure of Invention Aiming at the technical problems, the invention provides the infrared weak and small moving object edge detection method based on the frame difference local contrast, which is based on the fact that the detection of the weak and small infrared objects under the complex dynamic background is difficult to meet the requirements of the detection rate and the false alarm rate in the actual engineering, and effectively reduces the number of false alarms and improves the object detectability while ensuring the detection efficiency. The technical scheme adopted by the invention for solving the technical problems is as follows: The invention discloses an infrared weak and small moving object edge detection method based on frame difference local contrast, which comprises the following steps: step1, enhancing a target according to a three-frame difference method; Step 2, utilizing an improved local contrast computing method to inhibit background clutter and enhance the edge of a target; Step 3, target energy obtained by calculation through a central window is adopted to strengthen the target edge; And 4, performing threshold segmentation on the second characteristic diagram, and returning to the first differential characteristic diagram by using the coarse detection result to perform target positioning to obtain an accurate detection result. Further, the step 1 specifically includes: In the continuous three-frame image, the difference D Img3 between the second frame and the third frame and the difference D Img1 between the second frame and the first frame are calculated, and the first differential feature map DI Img for the second frame image detection is obtained again, the time domain noise is used to replace the spatial noise, and the contrast between the target and the background is enhanced. Further, the step 2 specifically includes: The infrared image comprises a background, an object and noise, wherein the background is in a gentle distribution state in space, is in a slowly-changing state in time, and the noise is in isolation in space and time. And (3) improving a local contrast algorithm, using a 2 multiplied by 2 sequencing filter to perform full-image sequencing filtering on the first differential feature map in the step (I), taking a gray intermediate value as gray estimation of a central target window, constructing an intermediate window to isolate a target background, performing local contrast calculation by utilizing the difference between the infrared weak target edge region and the adjacent background region, and inhibiting b