Study On Extraction And Applications Of Radar Target Physical Features

Radar is an all-weather,day/night detection device and has advantages of long range applications and optically opaque surfaces penetration.The advent of synthetic aperture radar and inverse synthetic aperture radar imaging technique make it possible to finer describe a target.Extracting useful information from SAR/ISAR image is very important in many military and civilian fields.The advent of polarimetric radar make it possible to extract physical scattering mechanism relevant polarimetric features from the polarimetric SAR/ISAR image.This paper studies on how to extract human-understandable physical features from target and how to improve the target detection performance by utilizing polarimetric features.The main content of this dissertation is summarized as the following four parts:The first part focuses on fast attribute scattering center(ASC)extraction based on sparse representation in image domain.Attribute scattering mode use a set of parameters to describe physical features including the backescattering intensity,location,shape,and orientation of an ASC.Extraction ASC from SAR/ISAR image can help SAR/ISAR image interpretation.The existing ASC extraction algorithms can be classified into two categories: category including image-domain algorithms and category including frequency-domain algorithms.Existing image-domain algorithms firstly segment the image.Therefor their extraction results depend on the segmentation results which are not precise when the target is intricacy.Frequencydomain algorithms can precisely extract ASCs,however they have high computation complexity and memory requirement.To tackle this problem,this dissertation proposes fast attribute scattering center(ASC)extraction algorithm based on sparse representation in image domain by utilizing properties of ASCs.The proposed algorithm has advantage of high accuracy as well as low computation complexity and memory requirement.The second part studies on solar panel size estimation and payload orientation estimation.Modern warfare is a high-tech information warfare and electronic warfare guided by information technology.The real-time monitoring and processing of battlefield dynamic information is an important factor related to the war.Solar panel size and payload orientation of space targets is an important information.This dissertation build a projection model of solar panel to the ISAR imaging plane,and define a concept of the distance between point cloud to a quadrangle.Base on the model and concept,this dissertation estimate the size and the attitude of the solar panel by using a series of ISAR images.This dissertation also proposes a orientation estimation method of the payload.In addition,the estimation error is analyzed and the conditions for the size estimation of the solar panel and the orientation estimation of payload is presented.The third part studies on extended geometrical perturbation based polarimetric SAR(PolSAR)image detectors.Geometrical perturbation based PolSAR detectors detect targets utilizing the coherent feature of two different backscattering maps.Existing geometrical perturbation based detectors includes single target detector,partial target detector,and geometrical perturbation-polarimetric notch filter.However,these methods model target and clutter as a vector which would file in the case target or clutter is complex.This dissertation model target or clutter as a subspace and proposes an extended geometrical perturbation PolSAR image detectors.Existing geometrical perturbation based PolSAR detector are special cases of proposed detectors which outperform existing detectors in the case clutter is sophisticate.The fourth part focuses on polarimetric contrast enhancement method based on minimum clutter-to-signal ratio(MCSR)subspace.This dissertation proposed a concept called MCSR subspace,and derive its solution algorithm.Projecting feature vectors extracted from polarimetric radar data can enhance the contrast between target and clutter and therefor improve the detection result.Compared with the existing optimization of polarimetric contrast enhancement(OPCE)and Generalized optimization of polarimetric contrast enhancement(GOPCE),the proposed method is more flexible and can be utilized in other fields.