Presegmentation-based adaptive CFAR detection for HFSWR

Common tasks of High-Frequency Surface Wave Radars (HFSWRs) are long-range ocean state monitoring and maritime surveillance with a strong focus on the detection of ships. Due to the heterogeneous background composed of sea-clutter and external noise the application of Constant False Alarm Rate (CFAR) algorithms with a single parameter set are likely to lead to a high probability of false alarm or poor detection performance. This paper is about adaptive CFAR with presegmentation, where the presegmentation is performed globally on each range-Doppler map and divides the detection background into external noise dominated regions and sea-clutter dominated regions. With this global knowledge it is possible to individually adapt the shape of the reference window for each Cell Under Test (CUT) to obtain homogeneous reference cells and avoid clutter-edges in the reference window. To further increase detection performance, the constant scale factor is chosen with respect to the current background. This enables detection of small targets in clutter while maintaining a low false-alarm rate for targets in external noise. To prevent the saturation of the tracker, a pretracker structure is presented which distinguishes between strong and weak detections and assigns priority to strong detections.