Research On Status Perception Technology For Moving Target In VideoSAR

Video Synthetic Aperture Radar(Video SAR),who introduces temporal dimension into SAR data processing,achieves dynamic monitoring and real-time reconnaissance via high framerate and high resolution imaging.Shadow-based moving target status perception technology in Video SAR is a research hotspot of current SAR field.Based on some mature algorithms for optical moving target detection and tracking,this novel technology comprehensively considers the characteristics of Video SAR system and corrects them pertinently.By detecting and tracking the shadow formed by the moving target in Video SAR image sequence,this technology can effectively compensate the system complexity and low azimuth positioning accuracy of the conventional GMTI(Ground Moving Target Identification)technology,holding the advantages of excellent positioning accuracy,high detection probability and low minimum detectable speed.However,due to the close relationship with target scattering characteristics,speckle noise,frame rate,geometric resolution,target size and target speed,shadow in Video SAR owns more severe instability problem compared with features of moving target in optical video.The loss of SCNR in shadow region occurs quite easily and thus makes shadow detection difficult to achieve.In a word,the unique imaging mechanism and image features of Video SAR increase the difficulty of moving target status perception,with the shortages of high false alarm rate,poor reality of measurement data and large speed estimation error.In view of the above mentioned problem,this thesis carries out a full discussion on Video SAR image sequence preprocessing,shadow-based moving target detection,Video SAR moving target tracking and optical-SAR video track fusion,establishing a complete implementation flow of Video SAR moving target status perception technology.All these efforts are expected to bring in some effort on enhancing our military’s battlefield situational awareness.The main contributions of this thesis are summarized as follows:1.The Video SAR image sequence preprocessing technique is studied.Firstly,based on the imaging principle of Video SAR system,the necessity of Video SAR image sequence preprocessing are analyzed,including registration and denoising.In the aspect of registration,the processing flow of SURF algorithm and its application problem in Video SAR are described.Then the idea of inter-frame accumulation is introduced and combined with SURF,forming a Video SAR image sequence registration algorithm based on inter-frame accumulation.In the aspect of denoising,the V-BM3 D algorithm is introduced and its performance pelter is also analyzed based on the inherent characteristics of the Video SAR.Then this video denoising algorithm is combined with temporal information.Finally,the performance of Video SAR preprocessing proposed in this thesis is verified by simulation and real data.2.The shadow-based Video SAR moving target detection technology is studied.Based on the preprocessing of Video SAR image sequence,the formation mechanism of moving target shadow in Video SAR is analyzed firstly,and the characteristics of shadow area and gray scale are mathematically derived.Then the moving target combining background difference and symmetric difference is proposed.This algorithm can accurately separate the moving target and the stationary background in the SAR video.Then,based on the characteristics of the amount of false alarms in the Video SAR and the false alarm spatial distribution in the edge region,the false alarm suppression algorithm based on gradient-weighted edge information is designed.It effectively suppresses a large number of false alarms in the edge region and minimizes the impact on detection performance.Finally,the effectiveness of the above algorithm is tested by Video SAR data.3.The Video SAR moving target tracking and multi-source video data track fusion technology are studied.Firstly,the preprocessing technique of multi-sensor measurement data is introduced as the foundation of track fusion.Then the Kalman filter is introduced and explain in detail.It is applied to the Video SAR moving target tracking technology.Finally,the multi-source data track fusion problem of optical and SAR sensor is explored.Combined with the moving target detection result of optical video of the same scene,the related research on multi-source heterogeneous sensor track fusion problem is carried out.Finally the above algorithm was verified via simulation data and real data.The research of this thesis covers the entire processing of Video SAR moving target status perception technology,and conducts a preliminary exploration of the problem of optical-SAR sensor track fusion.Such efforts effectively promote the popularization of this cutting-edge technology,and are expected to offer supports for the development and application of the future space-borne Video SAR system.