Eigenvalue/eigenvector-based serial decomposition of the polarimetric synthetic aperture radar coherency matrix

Abstract

Information loss and proliferation of parameters (PoP) in polarimetric synthetic aperture radar (PolSAR) eigenvalue/ eigenvector-based target decomposition had not been well treated until Paladini et al. recently devised a lossless and sufficient decomposition of the coherency matrix under circular polarisation basis, and a classification scheme based on the decomposed parameters was also proposed to validate the decomposed parameters. Inspired by the Paladini incoherent target decomposition, a new lossless and sufficient eigenvalue/eigenvector-based decomposition of the coherency matrix under Pauli basis is proposed. Compared to the coherency matrix under circular basis, the coherency matrix under Pauli basis is more frequently used because it is more convenient for interpreting the polarimetric scattering mechanism. The proposed decomposition not only deals with these two problems (information loss and PoP) well but also avoids the possible ambiguity of the α angle of the Paladini decomposition. In addition, both the two decompositions are serial or multiplicative decomposition, while other common target decompositions are parallel or additive decompositions. A further modified fine classification scheme is carried out to validate the decomposed parameters. Real PolSAR data are processed to validate the proposed decomposition.