A novel method for ship trajectory clustering
Helong Shen, Huang Tang, Yong Yin
Abstract
Different from trajectories in other fields, each ship trajectory has a unique direction. However, the traditional methods cannot be used to distinguish the trajectories in opposite directions. This paper presents an improved measurement with Cosine and Hausdorff distance to measure the difference between any two trajectories in different directions. In addition, to improve the efficiency of the clustering algorithm, a method named Course-Preserving Trajectory Simplification with Adaptive Compression Ratio (CPTS-ACR) has been proposed to simplify the trajectories. Ground-truth AIS data has been used to make a benchmark dataset. Experimental results show that the improved method in this paper has a better performance compared with some other existing methods.