Maritime Trajectory Clustering Method Based on Hausdorff Distance

In order to achieve the maritime intelligent navigation, it is necessary to discovery the historical navigation regularity for the vessels, so as to obtain the typical movement pattern of the ship sailing on the sea. Clustering method is an effective way to carry out statistical study on maritime traffic flow. This paper compared and analyzed the effect with different clustering methods applied to the maritime historical trajectory. Considering the structural features of the maritime trajectory, we discussed a method for maritime trajectory clustering based on Hausdorff distance and we also used the real Automatic Identification System (AIS) data to verify this method. Through the cluster analysis, the typical movement pattern of the ship is obtained. The experiment results show that the Hausdorff distance method can measure the similarity between maritime trajectories and has high accuracy. Finally, we compared the experiment result with the Density-Based Spatial Clustering of Application with Noise (DBSCAN) algorithm, and got better effect.