Ship behavior recognition based on infrared video analysis in a maritime environment

In order to improve the monitoring capability of officer on watch for the environment around the own ship, cover the shortage of RADAR and AIS in ship behavior recognition field, this paper proposes a ship behavior recognition algorithm based on video analysis. After analyzing the ship behavior we found that the silhouette's size and shape variation of ship in infrared image were related to its behavior over a period of time, that is to say when the distance between the own ship and the target ship has decreased during this period, its silhouette's size would increase. And when the target ship is heading toward the own ship directly and the distance between them has decreased (the DCPA (Distance to Closest Point of Approach) is very small, and the TCPA (Time to Closest Point of Approach) is decreasing), the silhouette's size of the target ship would enlarge, nevertheless, its silhouette's shape would have no significant changes. According to these characteristics of ship behavior, the height of target ship's silhouette is used to represent silhouette's size and the ratio between the width and the height of the target ship's silhouette is used to represent silhouette's shape in the ship behavior recognition algorithm we proposed. And then the least square method is used to fit the height and the ratio into straight line during this period respectively, whose slope reflects the changing trend and variation of target ship's silhouette. At last, the recognition algorithm was verified by analyzing infrared video in a maritime environment, which shows that the characteristic of ship behavior summarized by this paper is correct and the recognition algorithm based on them is feasible.