SAR image change detection based on Monte Carlo sampling

Abstract

ABSTRACTThe Kullback-Leibler divergence, as a distance measure between two probability density functions (PDFs), is used in many works for synthetic aperture radar (SAR) image change detection. However, the applicability depends largely on the underlying SAR image models. For instance, distribution has been demonstrated to be very flexible for modeling heterogeneous SAR images, the Kullback-Leibler divergence between two distributions is yet not analytically tractable. Therefore, in this article, we propose to use Monte Carlo sampling to approximate the Kullback-Leibler divergence between two distributions for SAR image change detection. The method of log-cumulants (MoLC) based on second kind statistics is applied for estimating the parameters in a distribution. Then a large number of independent and identically distributed samples are drawn from a distribution and used for the approximation. Through experiments, we compare this method with other state-of-the-art methods and demonstrate the competitive performance of the proposed method for SAR image change detection.