Research On Moving Target Detection Method Of Video SAR

Video Synthetic Aperture Radar(Video SAR)is a new SAR imaging mode.Because of its high frame rate imaging characteristics,it can effectively solve the traditional SAR ground moving target detection(GMTI)method's shortcomings.The new method provide a new solution to the detection of moving targets,so it has important research significance.In Video SAR moving target detection,an important method is to use the moving target's shadow and its state in successive frames to obtain the moving target's motion parameters.However,moving target shadows are difficult to distinguish from static target shadows in Video SAR images,thereby increasing the difficulty of moving target shadow detection.Therefore,it is of great significance to deeply study the theory and method of moving target detection based on shadow in Video SAR.This thesis focuses on the problems of Video SAR system moving target detection methods.The related work and progress achieved are as follows:1.First,the theoretical basis of Video SAR shadow detection method is studied.The formation conditions of moving target shadows in Video SAR images are analyzed,the mathematical model of moving target shadows is established,the theoretical formula of moving target shadow changes is derived,and the relationship between shadows of stationary targets and moving targets is analyzed,providing a theoretical basis for Video SAR moving target detection.2.The single-frame image shadow detection method(S-DT-OTSU)based on the Double Threshold OTSU(DT-OTSU)and the shadow tracking method based on the Kernel Correlation Filter(KCF)are studied.Video SAR moving target speed estimation method combined with DT-OTSU and KCF(S-DT-OTSU-KCF).The thesis gives the experimental results of simulating Video SAR image data.The results show that the method can detect the shadow of a moving target and estimate the speed of the moving target under the given effective shadow size range.3.Aiming at the shortcomings of the S-DT-OTSU method that requires effective shadow size range,the Video SAR moving target shadow detection method based on multi-frame image difference(M-DIFF)is studied.The thesis gives the experimental results of simulating Video SAR image data.The results show that compared with the S-DT-OTSU method,the M-DIFF method avoids the limitation of the shadow size range,effectively removes the shadow area of the stationary target,and reduces the false alarm rate.4.Aiming at the shortcomings of the S-DT-OTSU method requiring effective shadow size range and high false alarm rate,and the M-DIFF method high false alarm rate,a moving target shadow detection method based on convolutional neural network(CNN-MSD)is proposed.The thesis gives the experimental results of simulating Video SAR image data.The results show that this method uses the U-net network model,which not only avoids the limitation of the shadow size range in the S-DT-OTSU method,but also reduces the false alarm rate of the S-DT-OTSU method and the M-DIFF method.