Polarimetric SAR target decomposition based on nonlinear decorrelation

In remotely sensed Synthetic Aperture Radar (SAR) images, scattering from a target is often the result of a mixture of different scattering mechanisms. Fully polarimetric data offers the possibility to separate and to interpret them. To achieve this task, several target decomposition techniques have been proposed in the literature. In particular, the advantage of classical target decomposition techniques applied to fully polarimetric SAR data is due to the fact that by evaluating the scattering matrix, various scattering mechanisms and target properties can be identified. Aim of this paper is to evaluate a novel approach based on the use of Nonlinear Principal Component Analysis for the target decomposition. In fact, differently from classical target decomposition techniques, the proposed method is based on the decorrelation of the polarimetric SAR data to extract the inherent information content related to the different scattering mechanisms present in the image. An assessment of the effectiveness of the nonlinear principal component analysis method for target decomposition has been carried out by comparing it with the classical decompositions.