POLARIMETRIC DISCRIMINATORS FOR SAR IMAGES

A new method is developed to optimize the degree of polarization of a partially polarized wave reflected by a nonstationary scattering object. The method permits, for a scattered wave and a given target characterized by its Mueller matrix, analytic computation of the maximum and minimum values of the degree of polarization, and the corresponding transmitted polarizations. A procedure for the optimization of the scattered wave intensity is also proposed. The degree of polarization and the total scattered intensity extrema are then analyzed experimentally on JPL data. It is shown that several entities such as the received intensity extrema, the coefficient of variation, the fractional polarization and the span, which are currently used for target discrimination, can be deduced from combinations of the maximum and minimum values of the degree of polarization and the scattered wave intensity. Finally, a classification of the San Francisco image, based on these indices, is conducted for a better understanding of the specific physical meaning of each index.