A new clutter rejection method using a robust polarimetric CFAR-detector

Two methods for constructing robust polarimetric constant-false-alarm-rate (CFAR) detectors that use elements of the scattering matrix are discussed. Both methods use robust estimators to recognize outliers and exclude them from further calculations. The first method weighs each sample of the surrounding vectors, and vectors that appear to be outliers are weighted with lower values than the others. The second method uses cluster algorithms to arrange the data in different clusters; some clusters contain the outliers, and others contain observations assumed to come from the main body of the data. The detectors are intended to be used in multitarget and nonhomogeneous-clutter environments.<>