A polarimetric CFAR ship detector based on the joint probability function of simulated first-order Markov chains

In this work, a novel ship detection procedure is presented for radar signals whose backscattering power is of the same order while their polarimetric characteristics differ greatly. The stochastic nature of the polarimetric characteristics of the background (clutter) and the target signal (ship) can be adequately described by first-order Markov chains (FMCs). The proposed feature corresponds to the joint probability of sequential states of variable finite sequential segments of the FMCs. The classic Constant False Alarm Rate (CFAR) detection theory is adapted to the discrete space of the Markov experimental distributions of the proposed feature while the adaptive thresholding technique is overcome by considering a global model for describing the clutter. Performance assessment of the proposed detection procedure is carried out by means of three sets of FMCs each one corresponding to a clutter-target pair for different lengths of the finite FMC segments. The experimental results present high detection scores making the proposed detection procedure ideal for signals characterized by the Markov property. A comparative study of previous detection approaches has been implemented that shows the superiority of the proposed detection procedure.