Extraction of Ship Route Using Gaussian Process Regression for Passenger Ships

The extraction of passenger ship routes plays a crucial role in managing and regulating passenger ship traffic and ensuring safe navigation. This paper proposes a novel method for extracting passenger ship routes based on the Gaussian process regression (GPR) model. The method utilizes the Automatic Identification System (AIS) data to obtain the berthing trajectory segments based on the behavioral characteristics of ships. Next, the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) method is used to identify the berthing areas. The passenger ship trajectories are divided based on the identified berthing areas and are classified into different trajectory clusters based on their departure and arrival zones. The trajectory length and geographic coordinates (longitude and latitude) are modeled separately, and the GPR is employed to extract the trajectory centerline by aggregating the posterior distribution of the functions. The proposed method is evaluated by experimentation, which demonstrates its effectiveness in estimating the trajectory centerline more accurately. The research findings provide valuable support for designing and optimizing passenger ship routes and enhancing safety management.