Decoupling Between Different Polarization Channels of PolSAR Data

Purer contributions obtained from three diagonal elements of the coherency matrix are expected. This makes polarimetric synthetic aperture radar (PolSAR) easier to understand and to use in some applications, since the main information is aggregated into three mutually independent polarization channels. Therefore, we propose a method to decouple different polarization channels (DDPCs) in this letter. In such a case, all the nondiagonal elements of the coherency matrix are turned to zero, and three uncorrelated polarization channels are obtained. For this purpose, two rotation angles with physical meaning are proposed. They are responsible for rotating $T_{12}$ to zero. Radarsat-2 data are used to validate the proposed method. The results show that the proposed rotation angles are effective for land-cover classification. Furthermore, three polarization channels obtained from the DDPC method can mitigate the overestimation of volume scattering power for oriented buildings.