A novel polarimetric SAR ship detection method

This paper proposes a new ship detection method, which utilizes both the polarimetric statistical and scattering characteristic differences between ship targets and sea to improve the detection performance. Firstly, in order to enhance the ships in an unsupervised way with high signal to clutter ratio (SCR), a new metric is proposed based on the polarimetric scattering similarity, randomness and the intensity differences of ships and sea. Then grounded on the product model, the statistical modeling of the metric is presented with the radar cross section (RCS) modeled in the inverse gamma distribution. It assures the new metric can be used in clutters with different degree of homogeneity. Finally the effectiveness of the CFAR detection based on the statistical model is verified with RADARSAT-2 Polarimetric Synthetic Aperture Radar (POLSAR) data.