Mixture Generalized-Gamma Distribution for Sea-Clutter and its Parameter Estimation Method

Abstract

The paper proposes the mixture generalized gamma (G-gamma) model for sea clutter. Such distribution is useful, because the mixture gamma, the mixture Weibull and the mixture Nakagami distributions are special cases. A hybrid estimation method based on integer order moments and simplex search algorithm is derived in order to reduce the shape parameters finding complexity in four dimensions (4D) instead of 7D. To this end, a linear equation of second order is obtained to solve the probability variable as a function of the estimated shape parameters. The scale parameter is then calculated as a function of the first order moment and the estimated probability and shape parameters. The latter are computed by means of the recursive Least Square (LS) method with the simplex search algorithm. Modeling of IPIX (Intelligent PIxel X-bande radar) sea clutter is investigated using both single and mixture G-gamma distributions. The convergence of the proposed hybrid estimation approach is confirmed in terms of several scenes of IPIX database. It is also observed that the mixture G-gamma distribution exhibits the best fit to real data in most cases.