A New Decomposition of a POLSAR Coherency Matrix Using a Generalized Scattering Model

A new POLSAR model-based decomposition with a generalized scattering model is proposed. The generalized scattering model is built based on the particle cloud model adding the ellipticity angle. The model is highly flexible so that it represents the surface, double-bounce, volume, and helix scattering. Moreover, a uniform decomposition procedure to employ the model is proposed. For the first step, polarimetric coherency matrix is decomposed into two parts: 1) the generalized scattering model and 2) residue. Unknowns in the model are determined so that the resultant model matrix is subtracted, as much as possible, keeping the nonnegative eigenvalue condition in the residue. Then, the same operation is applied to the residue. By iteratively applying the process, the residue becomes null at last and the decomposed matrices perfectly fit the observed one. The procedure avoids branch conditions through the decomposition which often produce unreasonable decomposed power images. The proposed procedure is regarded as an extension of the eigenvalue-based decomposition considering the depolarization. The demonstration of the proposed decomposition shows that the results are considerably different from those of the eigenvalue-based one especially in forest area: more straightforwardly interpreted, since the model fits observed coherency matrix better. Furthermore, it is shown that the orientation angle randomness is physically important not only to measure the reliability of estimated orientation angles but also to estimate volume scattering power more accurately conventional model-based decomposition.