Research On Real-time Detection And Tracking Algorithm Of UAV In Video

It can be foreseen that small UAVs will play an important role in military and civilian areas in the future,and the UAV supervision problem will become more and more important,which brings a strong practical significance for the research of anti-UAV technology.Target detection technology and target tracking technology are important components of anti-UAV technology.They provide technical support for the following parts of target recognition and destruction(jamming).At present,video detection and tracking methods have the advantages of high visualization,high accuracy and low cost,and become one of the research hotspots.However,due to the complex outdoor scenes,the small speed and partial occlusion of unmanned aerial vehicle(UAV)target,it is very difficult to detect and track the target.This leads to the immaturity of the current visual detection and tracking technology,especially in the detection range and tracking accuracy.Aiming at the current problem,this paper designs a method of UAV intrusion detection and tracking in video.It applies the relevant methods in computer vision to detect and track UAV targets in video.The main research contents include intrusion detection algorithm and target tracking algorithm.The specific research contents are as follows:1.The common algorithms in moving target intrusion detection are studied.The basic principles,advantages and disadvantages of frame difference method,background difference method,optical flow method and some improved methods are briefly described.The problems in the direction of intrusion detection of UAV are summarized.In the tracking field,the basic principles of classical tracking algorithms such as Mean Shift algorithm,particle filter algorithm and correlation filter algorithm are briefly described,and the common problems of the above methods in tracking UAV are summarized;2.Aiming at the practical problems of UAV intrusion detection,such as too small target,complex background and more dynamic interference,this paper designs a detection method to solve the above problems.The method consists of three modules: inter-frame difference,background complexity analysis and improved saliency detection.Common methods This algorithm can effectively eliminate the dynamic interference of ground and air,improve the accuracy of detection,show better performance in complex scenes,and meet the requirements of real-time detection;3.The common problems in target UAV tracking are analyzed.Aiming at the unstable tracking situation of correlation filter tracking algorithm when the target moves faster and the target is smaller,a tracking method combining visual saliency detector and correlation filter algorithm is designed.This method shows better performance in more complex environment.Good performance;4.Detection distance is one of the important indicators of anti-UAV technology.The farther the detection distance is,the earlier the target will be found.In terms of algorithm,stable tracking can be achieved with very few pixels of tracking target imaging.For this reason,this paper designs a tracking method based on detection.Through the frame difference method,the target position can be positioned initially,and then stable tracking can be achieved by means of motion prediction and local precise positioning.