Change detection based GGR-GKIT on SAR amplitude image

In this paper, a novel approach to change detection from synthetic aperture radar (SAR) amplitude imagery is presented. In particular, a parametric pdf is proposed to model the ratio image, which is based on an efficient statistical model generalized Gamma distribution (GGD). The GGD forms a large variety of alternative distributions and is flexible to model the SAR images covering different kinds of surfaces in amplitude and intensity formats. In addition, a SAR specific parametric modeling approach and a suitable parameter-estimation based on the method of log-cumulants (MoLC) of the generalized Gamma ratio (GGR) distribution are adopted. Specifically, due to the considerable non-Gaussianity of SAR amplitude ratio data, the aforementioned parametric modeling approach for the ratio image is integrated into the generalized Kittler and Illingworth minimum-error thresholding (GKIT) process which allows one to take into account non-Gaussian data. Experimental results obtained confirm the effectiveness of the proposed approach.