Pietra index based processor for a heterogeneous Pareto background

This study addresses the problem of automatic target detection in a heterogeneous Pareto background. To achieve this, the Pietra index based and constant false alarm rate processor (CFAR) is conceived. Specifically, assuming a non-stationary Pareto background with the presence or not of any clutter edge or interfering targets, the Pietra index and the log geometric mean ratio statistic tests are concomitantly used to allow the proposed processor to switch dynamically to the appropriate detector; i.e. the geometric mean-CFAR, the greatest of-CFAR or the trimmed mean-CFAR, where all of these detectors assume a priori unknown scale parameter. That is, according to the outcomes of the Window Selection Probability, the background level is systematically estimated through the preselected detector. The detection performances of the proposed processor are assessed, via Monte Carlo simulations, in multiple target and clutter edge situations.