Trajectory similarity clustering based on multi-feature distance measurement

With the development of GPS-enabled devices, wireless communication and storage technologies, trajectories representing the mobility of moving objects are accumulated at an unprecedented pace. They contain a large amount of temporal and spatial semantic information. A great deal of valuable information can be obtained by mining and analyzing the trajectory dataset. Trajectory clustering is one of the simplest and most powerful methods to obtain knowledge from trajectory data, which is based on the similarity measure between trajectories. The existing similarity measurement methods cannot fully utilize the specific features of trajectory itself when measuring the distance between trajectories. In this paper, an enhanced trajectory model is proposed and a new trajectory clustering algorithm is presented based on multi-feature trajectory similarity measure, which can maximize the similarity of trajectories in the same cluster, and can be used to better serve for applications including traffic monitoring and road congestion prediction. Both the intuitive visualization presentation and the experimental results on synthetic and real trajectory datasets show that, compared to existing methods, the proposed approach improves the accuracy and efficiency of trajectory clustering.