Electromagnetic Feature Extraction Of Target From High Resolution SAR Imagery

As a method of acquiring information, Synthetic Aperture Radar (SAR) has already been an important tool for military reconnaissance and remote sensing due to its high resolution in both range and azimuth direction. The feature extraction is the foundation and the key element of high resolution SAR imagery interpretation. Electromagnetic feature is the inherent characteristic of SAR imagery, which is of great important for SAR imagery interpretation and target recognition. Scattering center is the essential electromagnetic scattering characteristic and a brief description of the radar target at high frequencies. Attributed scattering center feature, which can provide the target detailed information about its physical and geometric attribute, is the scattering center model applied to high resolution SAR imagery interpretation. Oriented to the application of SAR imagery interpretation, this paper mainly studies 2-D and 3-D attributed scattering center feature extraction of target from high resolution SAR imagery.Firstly, the concept of 2-D attributed scattering center feature, the parametric model and its normalization form are introduced, and the characteristic of the attributed scattering center model is analyzed by simulated data. Then, the RD-AML-CLEAN method for extracting scattering center feature from high resolution 2-D SAR imagery is studied. Finally the performance of the method is analyzed and evaluated qualitatively and quantitatively by simulated SAR imagery.Secondly, the paper studies the problem of electromagnetic feature extraction from measured high resolution SAR imagery and proposes the modified method. Parameter estimation and structure selection has been improved in this method. The validity of the modified method can be proved by the result of measured MSTAR (0.3m) and MiniSAR (0.1m) imagery data. Besides, on the basis of the modified RD-AML-CLEAN method, the attributed scattering center feature extracting software has been developed based on MATLAB GUI.Thirdly, based on 3-D attributed scattering center models of bistatic SAR, the paper do some groping researches on 3-D attributed scattering center feature extraction of target from high resolution SAR imagery. The third chapter introduces the 3-D attributed scattering center models of bistatic SAR. The validity of the parametric models can be proved by the results of simulated data. The fourth chapter analyses the parameter estimation of 3-D attributed scattering center feature extraction, and proposes the estimation method. Then, the result of stimulated experiment is given.