Ship Detection using Linear and Circular Range-Compressed Air-borne Radar Data

In the paper a novel real-time capable ship detection methodology for range-compressed (RC) airborne radar data is proposed. Ships are detected in the range-Doppler domain. The primary advantage of using range-Doppler domain is that ships moving with certain line-of-sight velocity are shifted to the exo-clutter region, thus improving their detection capability. Detection threshold is computed using constant false alarm rate (CFAR) based sea clutter models. Robust estimation of the detection threshold requires an accurate description of the background ocean training data. In the paper, a novel approach to extract reliable ocean training samples is discussed. In addition, different sea clutter models are investigated and compared to choose suitable models for the data. Real linearly and circularly acquired single-channel RC data from DLR's F-SAR system are used to verify the proposed methodology.