A CFAR Detection Algorithm for Weibull and Lognormal Clutter Mixture in SAR Images

In this paper, a Constant False Alarm Rate (CFAR) detector based on a Wei bull Lognormal Mixture Model (WLMM) for Synthetic Aperture Radar (SAR) clut-ter images is proposed. In doing this, the clutter parameters of the mixture and the corresponding percentiles are first estimated through the Maximum Likelihood Estimators (MLEs). Then, an adaptive threshold is set to maintain constant the Probability of False Alarm
$(P_{F^{\gamma}A})$
. The Simulated Annealing (SA) optimization algorithm is used to reduce the searching time in an efficient manner to approximate the global optimum percentiles. Application of the obtained CFAR framework on a real-world image, i.e., MSTAR BTR 60 (Moving and Stationary Target Acquisition and Recognition Bronetransporter), shows that an optimized weighting yields a good detection.