Fuzzy rule based classification of polarimetric SAR data

Abstract

We present a classification approach for full polarimetric SAR data based on Cloude's Decomposition Theorem. The approach is rule based, making use of knowledge of both the scattering properties contained in the entropy and α-angle values plus the backscatter intensities, which lie behind the first eigenvalue of the polarimetric coherency matrix. In order to overcome imprecise decision boundaries we make use of fuzzy logic. In a final step, the derived rulebase can be supervisedly optimized by a neuro-fuzzy approach. We show the performance of our approach on a data set taken by DLR's Experimental SAR (E-SAR) in L-band.