Vessel Trajectory Association Algorithm Based on Deep Learning Trajectory Similarity with Post-processing

With the continuous development of marine information technology in China, the volume of vessel trajectory data is growing rapidly. Trajectory association, as an important task in trajectory data mining and analysis, has become a significant research direction in vessel monitoring systems. To address the limitations of traditional trajectory association methods, which have high time complexity and low computational efficiency, as well as the insufficient capability of existing methods for vessel trajectory association, this paper proposes a Trajectory Association Framework based on Deep Learning Trajectory Similarity Calculation with Post-Processing (TADP). Experiments conducted on custom AIS and radar datasets show that the proposed method achieves a 7.5% improvement in mean rank for similar trajectory search on the AIS dataset and a 21.7% improvement on the radar dataset compared to T2vec. In robustness experiments, TADP outperforms other algorithms under different downsampling rates and distortion rates.