Shadow Based Moving Target Detection And Tracking Algorithm For Video SAR On Video SAR

Video SAR,a new concept of radar,effectively combines the all-time and all natural characteristics of traditional SAR and the imaging characteristics of high frame rate,which makes video SAR have far-reaching technical application potential and enhance the military precision strike ability and early warning ability for modern battlefield.The detection,location and tracking of moving target is an important research topic in the field of video SAR imaging,It has great research value and prospect.This paper mainly studies the shadow characteristics of moving target in the process of video SAR imaging,including moving target detection,tracking and positioning based on shadow.1.Aiming at the continuous imaging of the scene,this paper introduces the working mode of circular spotlight video SAR,analyzes the formation principle of the shadow of moving and stationary targets in the scene,and simulates the shadow of stationary and moving targets.In addition,convolution neural network is briefly introduced,and the principle of centernet network detection is analyzed from loss function and prediction network.2.Based on the moving target shadow in video SAR mode,the inter frame synthetic shadow detection is completed,and the depth network single frame detection method based on velocity partition prediction is proposed.Based on the shadow of moving target,the shadow of moving target is separated from the background by using the background composition of adjacent frames,and then the position detection of moving target is realized.The methods of inter frame registration and image gray matching are experimented and simulated,and the moving target detection experiment is carried out according to the video SAR imaging results published by SNL.In addition,from the experimental results,this paper discusses the problem that it is difficult to separate the shadow of high-speed moving target from the background based on multi frame fusion algorithm;Aiming at the problem of low detection rate of high-speed moving target,a single frame detection method based on velocity partition prediction is proposed,and a shadow velocity direction and size estimation algorithm based on Radon transform is proposed.Finally,the experimental test is carried out according to the video SAR imaging results published by SNL.3.The moving target tracking based on shadow is completed by Kalman filter method,and the moving target localization algorithm based on least square optimization is proposed.Kalman filter is used as the shadow moving target tracking algorithm,and the simulation experiment is carried out.The tracking experiments are carried out based on the Kalman filter and the video SAR imaging results published by SNL;Then,the positioning algorithm based on video SAR moving target is proposed,and the positioning algorithm is improved.The positioning accuracy is improved and the computational complexity is optimized based on the least square method.Finally,the optimization algorithm is verified by experiments based on inertial navigation and GPS positioning data.