Information fusion structure of VTS sensors for multi-target tracking of vessels

Abstract

Information fusion using VTS sensors of AIS, radar and camera are of great significance to the waterborne traffic supervision. Firstly, this paper invites shore-based CCTV cameras into detection and location of vessel targets combining with bounding boxes generated by deep-learning based detectors. Besides, this paper compares information fusion structure of central-level and track-level in simulated waterborne traffic scenario. Finally, this paper introduces a track selection method for sensors with large false alarm rate to obtain tracks with better performance on both fusion structure when fully considering strengths and weaknesses of all kinds of VTS sensors.

Keywords

• vts sensors
• information fusion
• central-level fusion
• tracklevel fusion