Improving the wishart synthetic aperture radar image classifications through deterministic simulated annealing

Abstract

This paper proposes the use of Deterministic Simulated Annealing (DSA) for Synthetic Aperture Radar (SAR) image classification for cluster refinement. We use the initial classification provided by the maximum- likelihood classifier based on the complex Wishart distribution that is then supplied to the DSA optimization approach. The goal is to improve the classification results obtained by the Wishart approach. The improvement is verified by computing a cluster separability coefficient. During the DSA optimization process, for each iteration and for each pixel, two consistency coefficients are computed taking into account two kinds of relations between the pixel under consideration and its neighbors. Based on these coefficients and on the information coming from the pixel itself, it is re-classified. Several experiments are carried out to verify that the proposed approach outperforms the Wishart strategy. We try to improve the classification results by considering the spatial influences received by a pixel through its neighbors. Finally, a link about the contribution of DSA to thematic mapping is also established.