An anti-interference method for capturing high-altitude parabolic objects by automatically filtering the shaking target area

Abstract

The present invention discloses an anti-interference method for capturing high-altitude parabolic objects by automatically filtering a swaying target area. The method comprises the following steps: S10. Identifying and dividing regions; S20. Applying a foreground modeling algorithm to the divided regions; S30. Automatically and cyclically updating and generating swaying regions and eliminating interfering objects based on the number and intersection of swaying regions; S40. Tracking the parabola; and S50. Generating a parabolic trajectory. By cyclically updating the swaying regions to generate and eliminate interfering objects, and further eliminating interference from interfering objects based on the parabolic trajectory, the method has substantial significance for capturing the parabolic trajectory.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and in particular to an anti-interference method for capturing high-altitude projectiles by automatically filtering a shaking target area. Background Art

[0002] Currently, traditional methods for detecting high-altitude parabolic objects have methods of drawing or generating building areas, but there are still high requirements for the installation location and content of the camera, and areas that are prone to shaking, such as trees or other vegetation, are not allowed to be blocked. Although some algorithms have proposed target filtering algorithms, the shaking of objects in actual applications will still seriously interfere with the parabolic tracking algorithm and thus affect the overall parabolic monitoring effect. Summary of the Invention

[0003] In order to overcome the deficiencies of the prior art, the present invention provides an anti-interference method for capturing high-altitude projectiles by automatically filtering a shaking target area.

[0004] The technical solution adopted by the present invention to solve the technical problem is: an anti-interference method for capturing high-altitude projectiles by automatically filtering a shaking target area, the improvement of which is that it includes the following steps:

[0005] S10. Access the camera address, extract a single frame image and draw the recognition area;

[0006] S20. Access the video stream, obtain each single frame in the video stream, crop the image to obtain the recognition area, apply the foreground modeling algorithm to the image, perform ecological transformation on the modeling mask, and generate the target recognition box;

[0007] S30. Determine whether a shaking region exists based on the number of target recognition frames. If a shaking region exists, initialize a shaking region mask and the original mask. Continuously count the number of foreground and background regions over a period of time based on the input mask to generate a shaking mask region. Perform morphological operations such as opening and closing operations on the generated shaking region. Merge shaking regions that are close to or enclosed by the shaking region. Output an automatically generated shaking region mask. Use the generated shaking region mask to mask the shaking region and repeat this step in a loop.

[0008] S40. Create a tracker for each target recognition frame, create a velocity model using Kalman filtering, and predict the target trajectory result;

[0009] S50. Use the predicted trajectory results and the actual trajectory results to perform IOU-based matching and update the tracker, and output the results. Determine whether it is an interference object based on the number of consecutive frames that have not been successfully matched, and filter out the target recognition frame determined to be an interference object;

[0010] S60. Draw the trajectory result of the target recognition box that meets the predicted trajectory result into the picture and the saved video for output.

[0011] In step S20 of the above technical solution, noise removal, blur removal, image binarization and image grayscale operations are performed on each acquired single-frame image.

[0012] In step S30 of the above technical solution, it is also necessary to filter the continuous shaking areas according to the intersection of the shaking areas to eliminate the interference of weeds or leaves.

[0013] The trajectory generation in step S60 in the above technical solution includes the following steps:

[0014] S601. Input tracking results;

[0015] S602. Establish a trajectory model for the same tracker;

[0016] S603. When the trajectory is not updated for a long time, it is defined as the end of the trajectory;

[0017] S604. Filtering insufficient number of trajectories and trajectories with wrong directions;

[0018] S605. Determine whether the trajectory meets the requirement of minimal shaking based on MSE, and filter out irregular trajectories such as those of mosquitoes and birds;

[0019] S606. Draw the result into a picture and a saved video for output.

[0020] In the above technical solution, the foreground algorithm in step S20 is the GMM or VIBE algorithm.

[0021] The beneficial effect of the present invention is that the present invention automatically generates an algorithm for areas prone to false alarms within an existing drawing area, such as leaves swaying in the wind, tree crowns, etc. By superimposing masks, it fundamentally solves a series of performance degradations caused by the swaying of trees or other objects in the algorithm area, which is of substantial significance for the capture of parabolic trajectories. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a flow chart of the anti-interference method for capturing high-altitude projectiles by automatically filtering a shaking target area according to the present invention. DETAILED DESCRIPTION

[0023] The present invention will be further described below with reference to the accompanying drawings and examples.

[0024] The following will clearly and completely describe the concept, specific structure and technical effects of the present invention in combination with the embodiments and drawings, so as to fully understand the purpose, characteristics and effects of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention. In addition, all the connection / connection relationships involved in the patent do not refer to the direct connection of components, but refer to the fact that a better connection structure can be formed by adding or reducing connection accessories according to the specific implementation situation. The various technical features in the invention can be combined interactively without conflicting with each other.

[0025] Reference Figure 1 As shown in the figure, the present invention provides an anti-interference method for capturing high-altitude projectiles by automatically filtering a shaking target area, comprising the following steps:

[0026] S10. Access the camera address, extract a single frame image and draw the recognition area, and divide the recognition area.

[0027] S20. Access the video stream, obtain each single-frame image in the video stream, perform noise removal, blur removal, image binarization, and image grayscale operations on each obtained single-frame image, crop the image to obtain the recognition area, apply the foreground modeling algorithm to the image, and use the VIBE algorithm to perform ecological transformation on the modeling mask to generate the target recognition box.

[0028] S30. Determine whether there is a shaking area based on the number of target recognition frames. If there is a shaking area, initialize a shaking area mask with the same size as the original mask to generate and record the existing shaking area. Continuously count the number of foreground and background for a period of time based on the input mask to generate a shaking mask area. Perform morphological operations such as opening and closing operations on the generated shaking area. Merge the shaking areas that are close to or wrapped. Filter the continuous shaking areas according to the intersection of the shaking areas to eliminate the interference of weeds or leaves. Output the automatically generated shaking area mask. Use the generated shaking area mask to shield the shaking area and then repeat this step in a loop. Circularly update the shaking area. The generated shaking area can be automatically updated in a loop.

[0029] S40. Create a tracker for each target recognition box, use Kalman filtering to create a velocity model, and predict the target trajectory result.

[0030] S50. Use the predicted trajectory results and the actual trajectory results to perform IOU-based matching and update the tracker, and output the results. Determine whether it is an interference object based on the number of consecutive frames that have not been successfully matched. If three consecutive frames have not been successfully matched, it is considered an interference object, and the target recognition frame determined to be an interference object is filtered out.

[0031] S60. Draw the trajectory result of the target recognition box that meets the predicted trajectory result into the picture and the saved video for output.

[0032] By superimposing masks, we can fundamentally solve the series of performance degradations caused by trees or other objects shaking in the algorithm area.

[0033] The trajectory generation in step S60 includes the following steps:

[0034] S601. Input tracking results.

[0035] S602. Establish a trajectory model for the same tracker.

[0036] S603: When the trajectory is not updated for a long time, it is defined as the trajectory ends.

[0037] S604. Filter out insufficient trajectories and trajectories with wrong directions.

[0038] S605: Determine whether the trajectory meets the requirement of minimal shaking based on MSE, and filter out irregular trajectories such as those of mosquitoes and flying birds.

[0039] S606. Draw the result into a picture and a saved video for output.

[0040] The tracking results are drawn into images and videos for output, while interference objects are further eliminated.

[0041] The above is a specific description of the preferred implementation of the present invention, but the invention is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A method for automatically filtering the shaking target area and capturing high-altitude parabolic objects with anti-interference, characterized in that The following steps are involved: S10. Access the camera address, extract a single frame image and draw the recognition area; S20. Access the video stream, obtain each single frame in the video stream, crop the image to obtain the recognition area, apply the foreground modeling algorithm to the image, perform ecological transformation on the modeling mask, and generate the target recognition box; S30. Determine whether a shaking region exists based on the number of target recognition frames. If a shaking region exists, initialize a shaking region mask and the original mask. Continuously count the number of foreground and background regions over a period of time based on the input mask to generate a shaking mask region. Perform morphological operations such as opening and closing operations on the generated shaking region. Merge shaking regions that are close to or enclosed by the shaking region. Output an automatically generated shaking region mask. Use the generated shaking region mask to mask the shaking region and repeat this step in a loop. S40. Create a tracker for each target recognition frame, create a velocity model using Kalman filtering, and predict the target trajectory result; S50. Use the predicted trajectory results and the actual trajectory results to perform IOU-based matching and update the tracker, and output the results. Determine whether it is an interference object based on the number of consecutive frames that have not been successfully matched, and filter out the target recognition frame determined to be an interference object; S60. Draw the trajectory result of the target recognition box that meets the predicted trajectory result into the picture and the saved video for output.

2. The method for automatically filtering a shaking target area and capturing high-altitude projectiles with anti-interference according to claim 1 is characterized in that: In step S20, noise removal, blur removal, image binarization and image grayscale operations are performed on each acquired single-frame image.

3. The method for automatically filtering a shaking target area and capturing high-altitude projectiles with anti-interference according to claim 1 is characterized in that: In step S30 , the continuous shaking areas need to be filtered according to the intersection of the shaking areas to eliminate interference from weeds or leaves.

4. The method for automatically filtering a shaking target area and capturing high-altitude projectiles with anti-interference according to claim 1 is characterized in that: The trajectory generation in step S60 includes the following steps: S601. Input tracking results; S602. Establish a trajectory model for the same tracker; S603. When the trajectory is not updated for a long time, it is defined as the end of the trajectory; S604. Filtering insufficient number of trajectories and trajectories with wrong directions; S605. Determine whether the trajectory meets the requirement of small shaking based on MSE, and filter out irregular trajectories of mosquitoes and birds; S606. Draw the result into a picture and a saved video for output.

5. The method for automatically filtering a shaking target area and capturing high-altitude projectiles with anti-interference according to claim 1 is characterized in that: The foreground algorithm in step S20 is the GMM or VIBE algorithm.

Images