Multi-Scale Ship Tracking Based On Maritime Monitoring Platform
Huanian Liao, Wei Wang, Shiqin Wang, Xin Xu · 2020
With the development of computer processing, network communication and other modern technology and equipment, the management of ships on the sea surface is becoming increasingly intelligent and automated. It is very important for navigation safety to obtain the behavior trajectory of the target ship. In this paper, a ship tracking method based on twin-neural network is proposed, which is mainly aimed at the deformation, blurring and blocking of ship movement at sea. The complex target ship in Marine environment around the real-time accurate detection of trace this research emphasis, In this paper, the ship tracking module based on correlation filtering method is firstly used to track the target ship, and then use the peak response than for water mist, and other vessels objects shade to what lead to the loss of the early warning, lost ship call based on the twin of the neural network for retrieving the lost ship detection module, continue to use the tracking module tracking target ship, the ship for a long time tracking under complex environment. Experimental results show that the tracking results of this algorithm are more accurate and real-time than those of typical algorithms.