Research On Intelligent Photoelectric Search And Tracking Technology Of Anti-UAV In Complex Background

With the change of international anti-terrorism and security forms,it is necessary to effectively prevent the "low,slow and small" targets from the air.Most of the air threats faced by urban air security are "low,slow and small" targets.The fixed wing target can fly at a speed of 30-50 m / s,with a large angular speed and maneuverability.In the case of target maneuvering,the high-precision intercepting system needs the laser ranging optical axis of the photoelectric tracking system to aim at the target in real time to obtain the target position information and estimate the target maneuvering motion parameters in real time,which puts forward high requirements for the tracking accuracy of the photoelectric tracking system.On the other hand,due to the large number of buildings and buildings in the urban environment,and the complex background,compared with the UAV target tracking in the normal clearance background,the target detection and image tracking ability in the complex background of the photoelectric search and tracking system also put forward new requirements.The intelligent photoelectric search and tracking system can realize the real-time search,capture and tracking of "low slow and small" targets in the air under the complex urban background,so as to provide the target motion parameters for the high-precision interception system.In view of the difficulties of target detection and high-precision tracking in complex background,this paper analyzes the imaging characteristics of target and complex background,puts forward the design scheme of multispectral multi-element detection optical system,extends the target information acquisition from a conventional single channel to multiple channels,so that the target and background can be distinguished in different wavebands.On the basis of multispectral imaging detection,this paper focuses on the target image tracking technology and high-precision servo tracking technology in complex background.On the basis of multispectral imaging detection,for the high-precision servo tracking technology in the case of target maneuvering,aiming at the different maneuverability and typical flight mode of various types of "low slow and small" targets,a neural network-based IMM Kalman filter feedforward compensation tracking method is proposed.In this method,the maneuvering characteristics of various types of targets are modeled and added to the IMM Kalman filter maneuvering model,and the neural network target recognition model is used to identify the air targets.According to the identified target types,the IMM Kalman filter parameters are automatically adjusted,so that the best estimation of the maneuvering characteristics of the targets is obtained by the filter.So it can provide accurate feedforward compensation control quantity for feedforward compensation control algorithm.The high precision servo control can ensure that the optical axis of the optical system can still be stably aligned with the target when the target is maneuvering,so that the ranging laser can continuously and real-time range the target.For the target image tracking technology in the complex urban background,a multi-mode composite TLD target tracking algorithm based on multi spectral detection is proposed.In practical system application,TLD algorithm has some disadvantages,such as time-consuming,easy to produce tracking drift and so on.Therefore,in order to obtain a real-time and stable tracking algorithm,this paper proposes an improved composite TLD target tracking algorithm.On the one hand,the image processing front-end first performs fusion processing on the acquired image,and the fused video frames are extracted(original 50 Hz,extracted 10Hz),then sent to the TLD target tracking algorithm,and the other is directly sent to the KCF target tracking algorithm,KCF calculation The method has high real-time performance and fast operation speed.Under normal operation,TLD algorithm will update KCF samples to make up for the situation that KCF algorithm can not adapt to target scale change and local occlusion,the outer layer adopts neural network target recognition technology based on prior information to recapture the target lost in the inner layer algorithm.In a word,the composite tracking algorithm complements the advantages of the three algorithms and improves the tracking stability and reliability.For the occlusion of UAV by buildings under the complex urban background,the prediction of the target's trajectory after entering the occlusion area is solved by using the IMM Kalman filter technology.When the UAV enters the occluded area,its trajectory prediction error increases with time.In a short time,the prediction accuracy of IMM Kalman filter is high.With the increase of time,the probability of various maneuvers of the target increases.An IMM Kalman filter-TLD target tracking algorithm against longstanding occlusion is proposed and verified via single machine test,performing simulation of multi-machine joint tracking.In this chapter,the algorithm adjusts the tracking gate according to the probability of the target's occurrence area,so that the target can re-enter the tracking field of view with a large probability after it leaves the occluded area.In this paper,on the basis of theoretical analysis of the above key technologies,relevant experiments are carried out to verify the effectiveness of the algorithm.For the tracking accuracy verification of maneuvering target,the target simulator is used to simulate various target maneuvers in the laboratory environment,and the photoelectric search tracking system is used to track and evaluate the tracking accuracy in real time.The accuracy of the IMM Kalman filter feedforward compensation controller based on neural network can be improved by more than three times compared with the conventional controller,and the actual system is verified in the field that the target maneuvering tracking accuracy is better than 0.5 mrad has verified the multi-mode composite TLD target tracking algorithm in the field environment.Compared with the conventional KCF or TLD algorithm,the average test accuracy of the video set with complex background is 0.9.In a single anti occlusion tracking test,the trajectory prediction error is reduced from 53 m to 15 m based on IMM Kalman filter.To improve the intelligent level,anti blocking ability and accurate tracking ability of the photoelectric search and tracking system is the development direction of the future anti UAV system with complex urban background.It is of great significance to study the key technologies of anti UAV,whether military or civil.