Analysis of PolSAR data of urban areas using time-frequency diversity

Abstract:
This article presents a new approach for the analysis of complex scenes and especially of urban areas, using PolSAR data. PolSAR information is considered in a time-frequency (TF) framework : signal is decomposed into sub-images corresponding to different angles of observation and frequency of illumination. The approach takes advantage of the diversity of the signal in the sub-images. Stationarity and coherency in the time-frequency domain are shown to provide complementary information, adapted to the case of man made environments. TF indicators, including polarimetric information, are derived for these two concepts. A TF based classification is proposed and applied to fully polarimetric SAR data acquired by ESAR sensor at L-band. Results show that the method is efficient to retrieve buildings location and that it supplies useful information for their characterisation.