TS-SSPD: A Method for Spatio-temporal Similarity Analysis Between Trajectories

Abstract

Trajectories representing the mobility of moving objects contain a great deal of spatio-temporal semantic information. Based on a measure of similarity defined between trajectories, trajectory clustering is one of the most simply and efficient ways to gain knowledge from trajectory data. Existing similarity measures do not take full advantage of spatio-temporal characteristics of the trajectory when measuring its distance. If the time dimension is not taken into account, the two trajectories in the opposite direction will be considered as similar. At the same time, in trajectory clustering, hotspot areas must also be determined according to time. In this paper, an improved SSPD distance TS-SSPD is defined, introducing the similarity of time dimension. A trajectory clustering algorithm based on TS-SSPD is also proposed. Compared this improved distance to the others according to their corresponding clustering results, the TS-SSPD improves the accuracy and efficiency of trajectory clustering.