Eigenvector analysis of polarimetric SAR data

Fully polarimetric SAR data analysis has found wide application for terrain classification, land-use, soil moisture and ground cover classification. New methods and algorithms continue to be developed and tested. One method that has gained great popularity is the Cloude-Pottier eigenvalue/eigenvector decomposition of coherency matrices [1]. The eigenvalue spectrum uniquely describes the scattering entropy, anisotropy and span (total power). The eigenvectors are employed in the calculation of averaged quantities. Typically though the eigenvector components have not been analyzed in much detail. Here we present a general analysis of the eigenvectors based on an earlier observation [2] that the eigenvectors may be parameterized by a set of rotations.