On a New ATR Tool for Coastal Surveillance Pulse Doppler Radars

We propose a new automatic target recognition (ATR) tool for coastal surveillance pulse-Doppler radars. The novelty of this tool consists of the new features for classification and a new classification algorithm which is performed by combining the predict models with M/N logic rule. Moreover, the detail design of ATR tool is also described. The proposed ATR tool is tested with real data for the purpose of fishing vessel recognition. The independent tests show that the proposed ATR tool has a better classification performance than that one given in some recent publications.