Effects of polarization on wide-angle SAR classification performance

Including polarization to circular synthetic aperture radar imagery increases the diversity of information as compared to non-polarized collections. The additional information improves classification performance in a civilian vehicle identification application. Effects of azimuth extent are also investigated. Radar imagery is represented as sets of attributed scattering centers, and vehicles are identified by a pyramid match hashing based, point pattern classifier. The experiments contrast classification performance in a variety of experiments with and without polarization.