Qualifying pixels for attributed scattering center extraction

Use of co- and cross-polarization phase history data has been shown to enhance feature extraction from synthetic aperture radar (SAR) data. Recent work with the spectrum-parted linked image test uses polarization with Fourier-based peak detection to qualify pixels for attributed scattering center (ASC) extraction from SAR phase history data. Extraction of closely spaced scattering centers suffers because only the dominant scatterer can be detected resulting in incomplete information about the object. We propose an ASC extraction method based on 2D Prony method and fully polarimetric SAR phase history similar to the SPLIT algorithm. In this work, we compare the signal parameter estimation performance when using a back-projection imaging operator to that when using 2D Prony method. We also examine the performance of feature estimation using experiments of primitive shapes that could represent a complicated feature needed for target classification.