Method for extracting miss distance of projectile drop point of staring type visible light photoelectric theodolite

Abstract

The invention relates to the field of image processing, and discloses a method for extracting the miss distance of a projectile drop point of a staring type visible light photoelectric theodolite. According to the implementation scheme, the method comprises the following steps: 1, determining a staring view field, acquiring a visible light image in real time by a photoelectric theodolite, and caching a background frame; 2, differentiating the current frame image obtained in real time and the background frame to obtain a differential image; 3, performing differential image target extraction based on multi-feature joint filtering on the differential image; 4, calculating coordinates of the center of mass and the lowest point of the target extracted in the current frame; and 5, judging whether the target centroid ordinate of the frame is the minimum value or not, and outputting the target lowest point miss distance when the target centroid ordinate is the minimum value. The target detection method based on multi-feature combined filtering is adopted, interference of strong light, noisy points, large-area background cloud and false targets is effectively removed, and the detection capacity of target drop points of visible light in a target range is improved.

Description

Technical Field

[0001] The invention relates to the field of image processing, and in particular to a method for extracting the amount of projectile drop point miss of a staring visible light photoelectric theodolite. Background Art

[0002] With the continuous development of digital image processing technology, visible light photoelectric theodolite target detection technology is also constantly developing and progressing. Photoelectric theodolite uses optoelectronic technology to measure the distance, height, angle and other targets. It is a very common photoelectric detection device and is widely used in military, aerospace, address, ocean and other fields. The commonly used detection method is moving target detection based on the frame difference method. This method can easily detect interfering targets, and multiple features need to be identified to lock the real target. Chinese patent 202211230068X discloses "A method for automatically interpreting images measured by a high-speed small target photoelectric theodolite", and the application publication number is CN115824165A. The interpretation method uses the secondary difference of the interpretation position of the previous few frames as a weight for correction, and combines the target image window position prediction method of the actual rotation angle of the main optical axis of the photoelectric theodolite. The prediction principle for the case of target missing and out of field of view is formulated, while ensuring the prediction efficiency, ensuring the accuracy and stability of the predicted position. By using the change angle of the main optical axis of the theodolite, a method for predicting the window position of the measurement image target recognition is proposed, which reduces the number of target searches and has a certain significance for improving the efficiency of automatic interpretation of high-speed small targets in the theodolite measurement image. However, this method cannot eliminate interfering targets and is easily affected by interfering targets and strong light, noise, and large-area moving background clouds. Summary of the invention

[0003] In order to solve the problem of interference of strong light, noise, large-area moving background clouds and false targets on target extraction, the present invention proposes a method for extracting the miss amount of projectile landing point of a staring visible light photoelectric theodolite. The interference of strong light, noise, large-area moving background clouds and false targets is effectively removed, and the detection capability of visible light target landing point in the shooting range is improved.

[0004] The present invention provides a method for extracting the amount of projectiles that miss the target by using a staring type visible light photoelectric theodolite, comprising the following steps:

[0005] 1) Point the photoelectric theodolite to the staring field, collect visible light images in real time, and cache background frames;

[0006] 2) The photoelectric theodolite acquires the visible light image in real time, and performs a difference between each frame of the visible light image and the background frame to obtain the difference image D i (x,y)(0<i≤N);

[0007] 3) Extract target from the differential image based on multi-feature joint filtering;

[0008] 3a) Setting the grayscale maximum threshold T2 and minimum threshold T1, binarizing the difference image, removing the glare, and generating a binary target candidate image B(x, y);

[0009]

[0010] 3b) marking the connected domain of the binary target candidate image B(x,y);

[0011] 3c) Setting a maximum threshold and a minimum threshold for the area of ​​the connected domain image, removing small noise points whose area is smaller than the minimum threshold and large background clouds whose area is larger than the maximum threshold, and obtaining the target in the frame image;

[0012] 4) Calculate the target centroid in the frame image and record the coordinates of the target lowest point;

[0013] 5) Determine and output the target in the frame image where the vertical coordinate of the target center of mass is the minimum value as the lowest point miss distance.

[0014] This technical solution can eliminate the interference of strong light, noise, large-area moving background clouds and pseudo targets on target extraction, and realize real-time detection of the impact point miss distance.

[0015] Preferably, in step 5), the method for judging and outputting the lowest miss amount is: starting from the second frame image, the center of mass ordinate of the current frame image target is compared with the center of mass ordinate of the previous frame image target; if the center of mass ordinate of the previous frame image target is smaller, the previous frame image is a landing point key frame, and the lowest miss amount of its target is calculated and output, and the iteration is terminated; if the center of mass ordinate of the current frame image target is smaller, the next frame image is selected as the current frame, and the comparison is continued until the center of mass ordinate of the previous frame image target is smaller, then the previous frame image is a landing point key frame, and the lowest miss amount of its target is calculated and output, and the iteration is terminated.

[0016] By adopting this technical solution, the lowest point miss distance can be accurately judged and output.

[0017] Preferably, in step 2), when the photoelectric theodolite acquires the visible light image, the area above the horizon is selected as the capture area, and the capture area and the background frame in each frame of the visible light image are differentiated.

[0018] By adopting this technical solution, interference in the area below the horizon can be eliminated, the amount of data processing can be reduced, and computing efficiency can be improved.

[0019] Preferably, when the number of pixels of the target image in the frame image is M*N, the maximum area threshold of the connected domain image is set to M*N+(100-1000), the minimum area threshold is set to 50-200, and the selected minimum area threshold is less than M*N.

[0020] This technical solution can effectively remove noise or background clouds.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] (1) Remove the interference of strong light, noise, large-area moving background clouds and pseudo targets on target extraction, improve the accuracy of target extraction, and improve the efficiency of extracting the projectile landing point miss distance.

[0023] (2) For a single target within a certain field of view, the automatic extraction of the impact point miss distance is realized, thus improving the ability to quickly report the target. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The present invention is a flow chart of a method for extracting the amount of projectile landing point miss. DETAILED DESCRIPTION

[0025] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] The present invention provides a method for extracting the amount of missed target from a projectile drop point of a staring visible light photoelectric theodolite, which is used to enable a user to perform target detection on a projectile drop image acquired in real time and extract the amount of missed target from the projectile drop point. The process of the method is as follows: Figure 1 As shown, the following steps are included:

[0027] 1) Point the photoelectric theodolite to the gaze field, collect visible light images in real time, and cache background frames; the gaze field refers to the position and range that the photoelectric theodolite can see. The gaze field can be determined by adjusting the direction and focal length of the photoelectric theodolite.

[0028] 2) The photoelectric theodolite acquires the visible light image in real time, and each frame of the visible light image is differentiated from the background frame to obtain the differential image. The differential image can be expressed by the following formula:

[0029] D i (x,y)(0<i≤N).

[0031] In addition, as a further improvement of the present invention, when the photoelectric theodolite acquires a visible light image, the area above the horizon is selected as the capture area, and the capture area in each frame of the visible light image is differentiated from the background frame to acquire a differential image expressed as described above. In this way, interference from the area below the horizon can be eliminated, the workload of data processing and calculation can be reduced, and the efficiency of data processing and calculation can be improved.

[0032] 3) performing differential image target extraction based on multi-feature joint filtering on the differential image; the differential image target extraction adopts the following method steps described in 3a)-3c):

[0033] 3a) Setting the grayscale maximum threshold T2 and minimum threshold T1, binarizing the difference image, removing the glare, and generating a binary target candidate image B(x, y);

[0034] The binary target candidate area image B(x, y) can be expressed by the following formula:

[0035]

[0036] 3b) marking the connected domain of the binary target candidate image B(x,y);

[0037] 3c) Setting a maximum area threshold and a minimum area threshold for the connected domain image, removing small noise points with an area smaller than the minimum area threshold and large background clouds with an area larger than the maximum area threshold, and obtaining the target in the frame image; in this step, when the number of pixels of the target image in the frame image is M*N, the maximum area threshold set for the connected domain image is M*N+(100-1000), and the minimum area threshold is 50-200. At the same time, ensure that the selected minimum threshold is smaller than the number of pixels of the target image is M*N.

[0038] 4) Calculate the target centroid in the frame image and record the coordinates of the target lowest point;

[0039] 5) Determine and output the target in the frame image with the minimum target center of mass ordinate as the lowest miss amount. In this step, the method for determining and outputting the lowest miss amount is: starting from the second frame image, compare the center of mass ordinate of the target in the current frame image with the center of mass ordinate of the target in the previous frame image. If the center of mass ordinate of the target in the previous frame image is smaller, the previous frame image is the landing point key frame, calculate and output its target lowest miss amount, and terminate the iteration; if the center of mass ordinate of the target in the current frame image is smaller, select the next frame image as the current frame, continue to compare, until the center of mass ordinate of the target in the previous frame image is smaller, then the previous frame image is the landing point key frame, calculate and output its target lowest miss amount, and terminate the iteration.

[0040] The present invention is described in detail below using a staring type visible light photoelectric theodolite projectile falling image with a resolution of 1280*1024.

[0041] Step 1: Determine the staring field of view. The photoelectric theodolite collects visible light images in real time and caches the background frames. Determine the focal length based on the target size, speed, and range to ensure that the field of view is large and the target has enough frames in the field of view. When the target size is 2m*1m, the focal length is selected as 50mm at a range of 1000m to ensure that the target pixel size does not exceed 100*20.

[0042] Step 2: Take the horizon as the dividing line and select the upper part as the capture area. Difference the real-time acquired current frame image and the background frame to obtain the differential image D i (x,y)(0<i≤N);

[0043] Step 3: Set the grayscale maximum threshold of 130 and the minimum threshold of 50 to binarize the difference image, remove the strong light, and generate the binary target candidate image B(x, y); mark the connected domain of the binary image B(x, y); set the maximum area threshold of 3000 and the minimum area threshold of 100 for the connected domain image to remove small noise points and large background clouds; of course, you can also select a maximum threshold of 2500 and a minimum threshold of 200, or a maximum threshold of 2100 and a minimum threshold of 150.

[0044] Step 4: Calculate the coordinates of the target center of mass and the lowest point of the target in the current frame;

[0045] Step 5: Determine whether the target mass center ordinate of the current frame is the minimum value, and output the target lowest point miss amount when the target mass center ordinate is the minimum value. Specifically, compare with the target mass center ordinate of the previous frame. If the current frame value is small, calculate the next frame mass center ordinate; if the current frame value is large, output the target lowest point coordinate of the previous frame.

Claims

1. A method for extracting the amount of projectiles that miss the target using a staring visible light photoelectric theodolite, characterized in that: The following steps are involved: 1) Point the photoelectric theodolite to the staring field, collect visible light images in real time, and cache background frames; 2) The photoelectric theodolite acquires the visible light image in real time, and performs a difference between each frame of the visible light image and the background frame to obtain the difference image D i (x,y)(0<i≤N); 3) Extract target from the differential image based on multi-feature joint filtering; 3a) Setting the grayscale maximum threshold T2 and minimum threshold T1, binarizing the difference image, removing the glare, and generating a binary target candidate image B(x, y); 3b) marking the connected domain of the binary target candidate image B(x,y); 3c) Setting a maximum threshold and a minimum threshold for the area of ​​the connected domain image, removing small noise points whose area is smaller than the minimum threshold and large background clouds whose area is larger than the maximum threshold, and obtaining the target in the frame image; 4) Calculate the target centroid in the frame image and record the coordinates of the target lowest point; 5) Determine and output the target in the frame image where the vertical coordinate of the target center of mass is the minimum value as the lowest point miss distance.

2. According to the method for extracting the amount of missed targets of projectiles falling from a staring type visible light photoelectric theodolite according to claim 1, it is characterized in that: In step 5), the method for judging and outputting the lowest miss amount is: starting from the second frame image, the centroid ordinate of the target in the current frame image is compared with the centroid ordinate of the target in the previous frame image; if the centroid ordinate of the target in the previous frame image is smaller, the previous frame image is a landing point key frame, and the lowest miss amount of its target is calculated and output, and the iteration is terminated; if the centroid ordinate of the target in the current frame image is smaller, the next frame image is selected as the current frame, and the comparison is continued until the centroid ordinate of the target in the previous frame image is smaller, then the previous frame image is a landing point key frame, and the lowest miss amount of its target is calculated and output, and the iteration is terminated.

3. According to claim 1 or 2, a method for extracting the amount of projectiles falling from the target using a staring type visible light photoelectric theodolite is characterized in that: In step 2), when the photoelectric theodolite acquires a visible light image, the area above the horizon is selected as the capture area, and the capture area in each frame of the visible light image is differentiated from the background frame.

4. A method for extracting the amount of projectiles missing the target from a staring type visible light photoelectric theodolite according to claim 1 or 2, characterized in that: When the number of pixels of the target image in the frame image is M*N, the maximum area threshold of the connected domain image is set to M*N+(100-1000), the minimum area threshold is 50-200, and the selected minimum area threshold is less than M*N.

Images