Research On Polarimetric SAR Data Decomposition Model

Polarimetric Synthetic Aperture Radar(Polarimetric Synthetic Aperture Radar)is a multiparameter,multi-channel active microwave imaging radar system,which makes use of different polarization combinations to measure ground targets with full polarization mode.It has become one of the hotspots in remote sensing field after rapid development in recent years.Polarimetric SAR target decomposition is the analysis of Polarimetric SAR data to extract the polarization scattering characteristics effectively in order to realize full polarization SAR data classification,detection and identification applications.Polarimetric SAR target decomposition has many schemes,there are two main categories.One is the polarimetric coherence target decomposition,which mainly uses the polarimetric scattering matrix to decompose the pure target;the other one is the polarimetric noncoherent target decomposition,which mainly uses the polarimetric covariance matrix or the polarimetric coherence matrix to decompose distributed target.Noncoherent target decomposition can be divided into noncoherent target decomposition based on Mueller matrix,noncoherent target decomposition based on eigenvalue decomposition and noncoherent target decomposition based on scattering model.In this paper,we study the noncoherent target decomposition method based on scattering model.There are three main problems in the decomposition methods based on the scattering model.The first one is that the overestimation of the volume scattering component in the man-made target areas because of the scattering model of the dihedral angle structures is considered as the volume scattering component in man-made areas.The second one is that there is occurrence of negative power in the decomposition methods.The third one is that there is no suitable volume scattering model to represent the volume scattering component.1.In the third chapter,a polarimetric SAR decomposition method based on generalized volume scattering model is proposed.This algorithm proposes two improvements for the original hybrid Freeman/Eigenvalue decomposition method in order to solve the problems of the volume scattering component overestimation and the lack of a suitable volume scattering model to represent the volume scattering component.The first one is that we use the rotated dihedral angle scattering model instead of the original volume scattering model when the polarization target is located in the urban areas or other man-made areas.This model can effectively reduce the proportion of volume scattering components in the manmade areas.The second one is that we use generalized volume scattering model,which is based on the adaptive parameters,when polarization target is located in the natural areas.It can effectively deal with various natural targets,but cannot effectively deal with man-made targets.The effectiveness of the new algorithm is verified by the comparison algorithm experiments.2.In the fourth chapter,a three-component decomposition method based on general scattering model is proposed.According to whether the polarization target is located in the regions where the volume scattering component is dominant or not,different polarimetric SAR decomposition scheme based on scattering model is carried out respectively.In the Cloude eigenvalue decomposition,a very important parameter which is the scattering entropy H is proposed,the larger the value of scattering entropy,the larger the depolarization effect of the target.When the value of scattering entropy exceeds a threshold,it can be regarded as a random scattering process,which is consistent with the polarization scattering mechanism of the forest regions and the regions where the volume scattering component is dominant.When the scattering entropy is smaller,the probability that the polarization target is deterministic scattering type is greater.Therefore,the threshold is determined by using the scattering entropy to distinguish between the region where the volume scattering is dominant and the region where the volume scattering is non-dominant.When the polarization target is located in the regions that the volume scattering is dominant,the polarimetric SAR target decomposition is carried out with using the Arii volume scattering model.When the polarization target is located in the regions that the surface scattering or double-bounce scattering is dominant,the generalized scattering model is used to decompose the polarization target.The effectiveness of the new algorithm is verified by comparing the algorithm experiments.3.In the third chapter,we introduce an Antropov generalized volume scattering model based on the reflection symmetry assumption.The model can characterize the natural regions effectively.In the fourth chapter,we introduce an Arii volume scattering model based on the non-reflective symmetry assumption.According to the research,it is found that the model can effectively retrieve the volume scattering characteristics of forest areas.