A hybrid technique for wake-based ship detection: Precise target localization by deterministic analysis of deep-learning segmented images

The paper shows a preliminary assessment of a novel hybrid technique for wake-based ship detection in SAR images. The combined approach leverages the more mature Radon-based wake detection in synergy with more recent deep learning object detection. The proposed methodology aims to handle the intrinsic limitations of both approaches. Results confirm that the diversity of the techniques can be deeply exploited to achieve a near-real time and reliable tool to improve the maritime situational awareness.