Ground Moving Targets Indication For Synthetic Aperture Radar Based On Shadow Detection And Sparse Representation

As a novel sensor with the advantage of all-weather, full-time, long-range,wide-field and high-resolution, synthetic aperture radar (SAR) has been widely used invariety fields, such as remote sensing, early warning, reconnoiter, guard against, etc.Ground moving targets indication has an important strategic significance on the hand ofthe observation between space and ground via SAR. Detecting moving target effectivelyand estimating the motion parameters and location accurately are the primary missions ofsurveillance radar. The study of effectively detecting, estimating and imaging groundmoving targets for single channel SAR has wide application future, without adding thehardware complexity.This thesis mainly studies the moving targets indication methods for single channelSAR based on image domain. The major devotement and innovatation of this thesis is asfollows:Firstly, explain the imaging mechanism and imaging algorithm in SAR and effectsof moving targets on the conventional SAR images, it analysis two moving targetsdetection approaches in detail, based on SAR image domain.Secondly, propose a moving target detection method, which is based on shadowdetection for single SAR imagery. And designed the whole flow chart of detectionprocess and elaborates on every detail. In this method, The shadow detection problem isdiscussed, the relation between the performance of shadow detection and the roadelectromagnetic scattering properties, clutter environment and noise intensity isproclaimed. The relation between moving target motion parameters and road isanalysised. Finally, combine shadow detection with road information, moving targets andmotion parameters can be confirmed.Finally, propose a moving target detection method, which is based on sparserepresentation. About this approach, the SAR/GMTI system model is built, and thecharacteristics of moving target is analysised, and then the ground moving target imagingmodel is achieved and the over-complete dictionary of targets sample images with different speeds is constructed. According to the corresponding coefficients, which arecalculated depend on the dictionary, moving targets and motion parameters can beconfirmed.