Three-Dimensional Feature Extraction And Modeling Of Target From High Resolution SAR Imagery

Synthetic Aperture Radar (SAR) has found success in a variety of applications to the economy and military of the country. It is a advanced and representative tool for suveying the earth all weather and all day at large standoff distances, which has the ability of penetrating the vegetation on the ground. Compared with the capability of acquiring data, high-resolution SAR imagery interpretation technology has not been developed enough. Feature extracting is the foundation and the key of SAR imagery interpratation. With the development of SAR system Three-dimensional feature extracting is a new research topic. Oriented to application of SAR imagery interpretation and target recognition, this thesis mainly researches the methods of 3-D geometric features extraction from SAR target and 3-D electromagnetic scattering feature modeling.Firstly, this thesis extracts 3-D geometric features from SAR target based on shadow. Shadow of target is ubiquitous in SAR imagery and the produce of shadow is relative to the height of target. The analysis results of the produce of shadow and the relation among the height of target, depression angle of radar and shadow show that it is possible to extract 3-D geometric features extraction from SAR target in high-resolution SAR imagery. The results of 3-D geometric features extraction based on single SAR image illustrate that the error is relatively large. To reduce the error the thesis proposes a method of 3-D geometric features extraction based on multi-SAR images by modifying the previous single SAR image based method.Secondly, this thesis models the 3-D electromagnetic scattering of target. Scattering center is the essential scattering characteristic of target in optical region. It is an important step to model the scattering center of targets. This thesis develops the attributed scattering center models by combining physical optics(PO) and uniform theory of diffraction (UTD) planar scattering solutions to approximate 3-D scattering responses of targets and analyses the rotated scattering mechanisms and polarization response. Validation results of the models using high frequency electromagnetic field compution software show that the proposed models capture well the mainlobe responses of target. Furthermore, validation of the model has been carried through by RCS analysis and 3-D SAR imagery simulation. Finally feature extracting method of 3-D scattering center is explored.