Modelling of sea clutter with Gaussian mixtures and estimation of the clutter parameter

In order to model sea clutter, we propose to use a generalized distribution which is valid for different background statistics, like low and high sea states. The envelope detector´s input statistics are assumed to be a Gaussian-mixture (GM). A new background probability density function (pdf) resembles Rician mixtures. In order to detect targets using adaptive techniques, clutter background statistics should be known a priori. In practice, however, background statistics are typically not known a priori. Thus, a joint estimation and detection process is employed whereby the background statistics are estimated before target detection. We estimate the parameters using a maximum likelihood (ML) based method from the output of the envelope detector. Performance analysis is presented using real sea clutter data.