Multi-granularity Visualization of Trajectory Clusters Using Sub-trajectory Clustering

Abstract

With the surging of the requirements of location-based services, mining various interesting patterns from the spatial data becomes more and more important. In this paper, we propose an approach for visualizing the trajectory clustering results based on sub-trajectory clusters discovered from large-scale trajectory data. At first, we segment each trajectory into a set of sub-trajectories by detecting its corner points. And then, we choose Fréchet distance to compute the similarity between sub-trajectories, and use a density-based clustering method to cluster sub-trajectories and get an augmented order of the sub-trajectories. The visualization method can support multi-granularity views of the generated sub-trajectory clusters. Experiments have demonstrated the applicability and benefits of the proposed approach.