Research On Target Detection And Tracking Technology For Anti-UAV System

Anti-UAV(Anti-Unmanned Aerial Vehicle)systems are widely used in counter-terrorism,air traffic control and privacy security In existing anti-UAV systems,the demands of defense and attack are seriously concerned.Unfortunately,the systems are easily detected during moving warfare because of its huge size,and occurred high false alarm rate under occlusion.In this paper,a miniaturized and high-precision anti-UAV detection and tracking system is developed to meet the requirements of local low-altitude defense.Target detection approach combined with visual attention mechanism is developed for small four-rotor UAV.Target tracking and trajectory prediction algorithm are designed for leafage backgrounds,and the performances of the system are verified by DaVinci hardware platform.The main contents of the paper include four aspects:1)The photoelectric imaging technology scheme is presented.The devices selection of the detection system is completed,which is based on analyzing the operating distance of the detector,the relationship between pixel size and resolution of the image sensor,the imaging process of target in optical system,and the spectral response characteristics of the detector.And the photoelectric imaging proposal is established for low altitude anti-UAV detection and tracking system.2)Unmanned aerial vehicle target detection method is built.By comparing three types classical saliency detection models,the color space model and characteristics of wavelet function are discussed.The method of extracting salient regions of targets using wavelet analysis is improved based on considering the local and global contrast characteristics of the image,The experiment results show that the precision,recall and F-measure of sym2 wavelet saliency detection method in XYZ color space are higher than those of the other five wavelets.3)The target tracking and trajectory prediction algorithms are designed.Based on filtering theory and occlusion discrimination strategy,the spatio-temporal context tracking algorithm integrated with Kalman filter is introduced to solve the drift problem,which is caused by global occlusion in leaf cluster scenes.The experiments proven that the average success rate of the proposed algorithm is 14.9%higher and the accuracy of center position error is improved by 6.8 pixels than the original algorithm.4)The embedded algorithm transplantation is exhibited.The running environment of the system is built,which is based on the software architecture of DM6446 DaVinci platform.According to the algorithm development processes,the Codec package,DSP server and Codec engine are configured and generated respectively.What’s more,the application program of the target tracking algorithm on the basis of the spatio-temporal context is developed.The results show that the processing speed of embedded target tracking system is at least 13 frames per second.The experiments verified that the anti-UAV target detection and tracking system constructed in this paper realizes the miniaturizated design,and achieves the expected goal of improving the tracking success rate and target positioning accuracy,which plays a significance role in the low-altitude defense.