Polarimetric synthetic aperture radar image unsupervised classification method based on artificial immune system

An unsupervised classification method based on the H/alpha classifier and artificial immune system (AIS) is proposed to overcome the inefficiencies that arise when traditional classification methods deal with polarimetric synthetic aperture radar (PolSAR) data having large numbers of overlapping pixels and excess polarimetric information. The method is composed of two steps. First, Cloude-Pottier decomposition is used to obtain the entropy H and the scattering angle alpha. The classification result based on the H/alpha plane is used to initialize the AIS algorithm. Second, to obtain accurate results, the AIS clonal selection algorithm is used to perform an iterative calculation. As a self-organizing, self-recognizing, and self-optimizing algorithm, the AIS is able to obtain a global optimal solution and better classification results by making use of both the scattering mechanism of ground features and polarimetric scattering characteristics. The effectiveness and feasibility of this method are demonstrated by experiments using a NASA-JPL PolSAR image and a high-resolution PolSAR image of Lingshui autonomous county in Hainan Province. (C) The Authors. Published by SPIE under a Creative Commons Attribution 3.0 Unported License.