Independent Component Analysis of POLSAR Images. Relative Newton-Based Approach

We propose here a new method for POLSAR image analysis. The method is based on a new PCA-ICA model in which the relative Newton-based approach for performing ICA is developed. The basic idea of ICA with relative Newton method consists in approximating the negentropy by taking account of the orthogonality constraint of the extracted components. This concept is recognized for its robustness and gives consequently very good theoretical results. The approach is well justified from the mathematical point of view. However, its implementation requires being more flexible because of the number of the estimated parameters. The purpose of this paper is to try to open new issues, in future research, in the concern of working out a new method for SAR image analysis that accumulate the advantages of the proposed method while avoiding its disadvantages.