The multilinear compound Gaussian distribution

Abstract

Abstract-We introduce a novel generalization of the compound Gaussian (CG) (

A nonlinear compound representation of sea clutter

“…We further point out in this connection that since CG model has been tremendously useful in image processing for applications such as denoising, we should expect similar applicability of the NCG model beyond the modeling of sea-clutter [1]. Thus NCG is a general model that could be used a building block for more complex probabilistic models of both sea-clutter and scene structure.…”

“…Matrix C , C , … , C T A is assumed to be invertible and thus determines the interaction between the various Gaussian sources. Vector , , … , is an instantiation of random variable of the NCG model corresponding to (1). The constants and play an important role in determining the properties of the non-linear transformation , as we show in the next section.…”

“…We introduce a novel generalization of the CG model called NCG [1] (nonlinear compound Gaussian) distribution which allows us to extend the capabilities of the CG model to better accommodate the variabilities in modeling requirements such as above. We give a self-contained description of this model together with its theoretical properties (in Section 2) and algorithmic details of its parameter estimation (in Section 3).…”