Statistical modeling of a joint metric of SAR interferogram's magnitude and phase using Fisher distribution for texture

Abstract

In this paper, aiming at the precise description of a recently proposed joint metric of SAR interferogram's magnitude and phase (IMP), especially for high resolution interferometric SAR images, a novel distribution denoted as S1 distribution is proposed. At first, under the framework of product model, the Fisher distribution is introduced for modeling the textural component and the closed-form PDF of the novel distribution is presented. The proposed distribution generalizes the existing S0 distribution which models the textural component by the inverse gamma distribution. Then, based on the Mellin transform, a parameter estimator of the proposed model is derived. Finally, the experimental results on real data have validated the efficiency of the proposed distribution and the parameter estimation.