A Feature-based Ship Detection Method for Compact Polarization SAR Image

Inhalt:
Compact polarimetric SAR (CP SAR) has triggered great consens in ocean maritime monitoring recently for containing abundant scattering information and the wide swath. In this paper, we propose a new feature-based method for ship detection using CP SAR. Four scattering parameters are combined to form a feature vector after a detection performance assessment: degree of polarization, entropy, scattering angle and the miner component of coherence matrix. Dual circular polarimetric (DCP) data simulated by Radarsat-2 Fine Quad mode images are utilized in this research. Compared with CFAR results of quad-polarimetric PWF, HH, span of DCP data and double-bounce component of m-chi decomposition, the new feature shows great capability in detecting very weak targets meanwhile reducing false alarms.