Model for non-rayleigh clutter amplitudes using compound inverse gaussian distribution: an experimental analysis

Abstract

A statistical model for high-resolution sea clutter, which we have called the compound inverse Gaussian (CIG) distribution, is proposed. The model is a mixture of the Rayleigh distribution and the inverse Gaussian (IG) distribution to model the speckle and the texture components, respectively. The CIG probability density function (pdf) is generalized to account for the additive thermal noise to achieve a good match to real data. The overall pdf is given in an integral form as a function of three parameters which are estimated from the recorded data based on the parametric curve-fitting estimation (PCFE) method of the complementary cumulative distributed function (CCDF). The Nelder-Mead (N-M) simplex algorithm is used to provide the best estimates of the pdf parameters. Using the IPIX backscatter database, the fitting of the CIG pdfs and the cumulative distributed functions (cdfs) are assessed and compared with the fitted Weibull, log-normal, Rician inverse Gaussian (RiIG), K plus noise, compound log-normal (CLN), and Pareto plus noise distributions. Experimental fitting results show that the sea-clutter amplitudes obey the proposed CIG model in most cases.