A noncentral and non-gaussian probability model for SAR data

Abstract

A general compound statistical model for coherent imaging is developed and tested on single-channel Synthetic Aperture Radar (SAR) data. In this formulation, coherent scattering is taken into consideration and the texture is modeled using an Inverse Gaussian distribution. Parameter estimation is conducted via an Expectation Maximization (EM) scheme. A Maximum a Posteriori (MAP) speckle filter based on this model is also implemented. The filter shows good smoothing capabilities and preserves details in the selected scene, showing promise for target-detection applications.