Vessel detection and classification fusing radar and vision data

Unmanned Surface Vehicles (USV) involves vehicle structure, computer science, robotic vision, automatic control and other areas, so the research on USV is significant in both theoretical and practical aspects. Vessel detection and recognition and tracking is mainly discussed in this paper by fusing data coming from a variety of sensors: a single camera, a radar, an inertial sensor. The vision is used to find the true targets detected by the radar and to discard those who might be false positives by the method of combining offline SVM classifier, saliency detection and online compressive tracking. Additionally, heading reference system (HRS) is used for compensating the angle deviation caused by the shaking of USV due to the fluctuation of waves. The true targets will be considered by the comprehensive decision of SVM classifier and radar cross-validation results and saliency detection validation radar results. Experiments confirm the effectiveness of the proposed architecture.