Classification of Polarimetric SAR Images Based on Optimum SVMs Classifier Using Bees Algorithm

Because of Polarimetric Synthetic Aperture Radar (PolSAR) contains the different features which relate to the physical properties of the terrain in unique ways, polarimetric imagery provides an efficient tool for the classification of the complex geographical areas. Support Vector Machines (SVMs) are particularly attractive in the remote sensing field due to their ability to handle the nonlinear classifier problem in high dimensional feature space. However, they also suffer from optimum SVMs parameters assignment and optimum feature subset selection issues that can significantly affect on the obtained results. In optimization of SVMs parameters and feature space, traditional optimization algorithms usually trap in local optimum. Thus, it is inevitable to apply metaheuristic optimization algorithms to obtain global optimum solution. As results, the superior performance of SVMs achieved by simultaneously optimization of SVMs parameters and input feature subset on Polarimetric imagery are demonstrated. Keywords—Polarimetric Image, Support Vector Machines (SVMs), Feature Selection, Model Selection, Bees Algorithm.