A Study Of Moving Target Detection And Tracking Techniques With Video SAR

Moving target detection and tracking technique is one of the most important research topics in the field of synthetic aperture radar(SAR).A ground moving target indicator(GMTI)is designed to detect moving targets by examining the target's reflected radar echoes.Compared with the single-channel system,multi-channel SAR systems are more widely used because of their better performance in clutter suppression.However,there are still many factors that impact the system performance of multi-channel SAR systems,such as channel mismatch,undulating topographic elevation,and inevitable “blind speed”,etc..In 2003,U.S.Sandia National Laboratories proposed a new method,video SAR imaging to realize continuous high-resolution video surveillance on a scene by sequentially displaying its SAR images.In the video SAR imagery sequences,a moving target leaves dynamic shadows in its real locations,which visually reflects precise state of the moving target so that we may detect and track the moving target through the shadow detection and tracking.But there is also a limitation on its detectable speed to form a moving target's shadow.Target tracking technique is used to obtain the trajectory of a moving target,and to estimate and predict target's motion parameters.The difficulty in target tracking technique is to accurately track maneuvering targets under diverse applications and uncertain targets.This thesis will deal with these issues on the moving target detection and tracking in a video SAR system.Main contributions and progresses are summarized as follows.1.Investigated the shadow detection approach for detecting slow-moving targets in SAR images.The conditions and practical factors for a moving target to form a recognizable shadow in a SAR image are analyzed.Taking the video SAR imaging data published by Sandia National Laboratories as an example,this effort successfully extracted the shadows of a moving target with unknown size and uncertain initiate position by image registration,background extraction,noise reduction,and threshold segmentation.2.To solve the problems such as inaccurate image registration and false alarm induced by static tall target etc.,we proposed an improved approach that jointly processes multi-frame images and utilizes the fact that the shadow of a stationary target is closely accompanied by its image.The test results on Sandia's experimental data show that our improved approach significantly reduces the missed alarm rate.3.Investigated the tracking methods for multiple targets.This effort analyzed the performances of typical algorithms for track initiation,data association and tracking filter.It is found that a compound algorithm that selects an appropriate algorithm on each stage of initiation,data association,and tracking,associated with our improved shadow detection method as mentioned above,is very effective in tracking multiple maneuvering targets.The simulation results validated the analysis.4.To reduce the sensitivity of the multi-channel GMTI to the channel mismatch and the effect of clutter on the interferometric phase,we introduce a three-channel algorithm combining along-track interferometry(ATI)and the displaced phase center antenna(DPCA)techniques to achieve both clutter cancellation and moving target detection.The obtained interferometric phase can be used to estimate the radial velocity of the target,and the target's geo-location with the help of the estimated Doppler parameter.5.Performed a Video SAR simulation experiment.Applying our shadow-based detection approach and the three-channel DPCA/ATI algorithm to the simulated SAR data,we have successfully detected both slow and fast moving targets in a complex environment.The simulation results validate the complementarity and effectiveness of these two approaches for the moving target detection and tracking in video SAR.