The Generalized Gamma Model has as special cases the Rayleigh, Weibull and Lognormal models. It also closely approximates the K-pdf model. Radar Clutter is often approximated in one of these forms. It is therefore quite useful to develop CFAR (Constant False Alarm Rate) detectors that perform well under this clutter model. In this paper, a Maximum Likelihood Generalized Gamma (MLGG) CFAR detector has been developed. This MLGG detector uses the Maximum Likelihood Equations, both locally and globally, in order to estimate the parameters of the Generalized Gamma clutter. These estimated parameters are then used to estimate the local mean of the detector. The mean of the local CFAR window is then taken as the first moment of the Generalized Gamma distribution evaluated with the estimated parameters. In the examples it is shown that in homogeneous Generalized Gamma clutter, with point targets, the MLGG detector outperforms our standard test detectors, Cell Averager, Ordered Statistic and Optimized Weibull.
展开▼
