Polarimetric decompostion based on partcle swarm optimization and its data analysis

This paper deals with a polarimetric decomposition considering azimuth rotation based on evolutionary algorithm. Polarimetric data in urban and mountain areas are affected by azimuth rotation. Thus, the decomposition results derived by same target with different azimuth rotation angle are changed. Some researchers examined to compensate this rotation by mathematical inverse rotation. In this report, the differences of covariance matrix before and after the compensation are used to estimate the contributions of three decomposition components. An evolutional algorithm to determine the contributions is a particle swarm optimization (PSO) which is one of the evolutional algorithms. We apply the proposed method to ALOS/PALSAR data and compare the results between four component decomposition method and proposed method.