Enhanced target characterization and improved scattering power decompositions using the optimized coherency matrix from full-polarimetric SAR data

Abstract

ABSTRACTIn this article, a methodology is proposed for enhanced target characterization and improvement in scattering powers from model-based decomposition using full Polarimetric Synthetic Aperture Radar (PolSAR) data. The novelty of this approach is established by utilizing the degree of polarization (DOP) from full PolSAR data. The optimum degree of polarization (DOPopt) is obtained by maximizing the DOP estimated from each of the four Mueller matrices corresponding to the original 3 × 3 coherency matrix and the three unitary transformed (one real and two complex) coherency matrices, respectively. The optimized coherency matrix which corresponds to the is then used for the enhanced target detection using a polarimetric similarity measure. It is observed that the optimized coherency matrix corresponding to a target is much closer to an elementary target (i.e., either dihedral, trihedral or helix) with a canonical scattering mechanism than the original coherency matrix. Furthermore, it can be seen that the scattering powers obtained from the Freeman–Durden-model-based polarimetric decomposition technique using this optimized coherency matrix shows good improvement than using the original coherency matrix. The double-bounce power over the urban areas has increased for the L-band Airborne Synthetic Aperture Radar (AIRSAR) San Francisco data and for the L-band Advanced Land Observing Satellite 2 (ALOS2) Kyoto, Japan data compared to the original PolSAR data. Negative powers have also decreased by the use of this novel technique.