Ship trajectory segmentation by movement states while addressing uncertainty and sparsity

The trajectory collected by AIS is an essential data source for various types of ship analysis, and segmenting a single trajectory into several more miniature and uniform segments helps extract valuable insights. However, most current trajectory segmentation methods are calibrated for high-frequency trajectories, which cannot adapt well to the uncertainty and sparsity of ship trajectories, indirectly affecting the accuracy of the trajectory analysis results. In light of this, a segmentation method is proposed to segment ship trajectories into continuous sequences of stop or move episodes without associating external data. Firstly, the uncertainty is eliminated by the glitch and jitter rectification methods, which correct finer noise while moving and merge jitter points when approaching the port. Secondly, to solve the sparsity, a ship trajectory interpolation method is designed for the long-term signal loss intervals associated with the movement states. Upon interpolated trajectory, an improved trajectory segmentation algorithm based on DBSCAN is proposed to cluster adjacent trajectory points with different sampling frequencies based on spatial–temporal characteristics and movement states, making the segmentation applicable to sparse ship trajectories. In the experimental stage, a publicly available segmented dataset is provided, including 100 randomly selected tanker and cargo ship trajectories, which allows ship trajectory segmentation to accommodate uncertainty and sparsity. The results show that the proposed method can effectively deal with the uncertainty and sparsity of ship trajectories and is robust to ship types, verifying its effectiveness and applicability. •A robust uncertainty correction algorithm is proposed, which eliminates uncertainty through glitch and jitter rectification.•A ship trajectory interpolation method for the long-term signal loss interval related to the movement states is designed.•An improved trajectory segmentation algorithm based on DBSCAN is proposed to cluster adjacent trajectory points with different sampling frequencies based on spatio-temporal characteristics.•A publicly available segmented dataset is provided, which makes up for the shortage of the current ship trajectory segmentation datasets with uncertainty and sparsity.