TCMVS: A Novel Trajectory Clustering Technique Based on Multi-View Similarity

The analysis of moving entities "trajectories" is an important task in different application domains, since it enables the analyst to design, evaluate and optimize navigation spaces. Trajectory clustering is aimed at identifying the objects moving in similar paths and it helps the analysis and obtaining of efficient patterns. Since clustering depends mainly on similarity, the computing similarity between trajectories is an equally important task. For defining the similarity between two trajectories, one needs to consider both the movement and the speed (i.e., the location and time) of the objects, along with the semantic features that may vary. Traditional similarity measures are based on a single viewpoint that cannot explore novel possibilities. Hence, this paper proposes a novel approach, i.e., multi viewpoint similarity measure for clustering trajectories and presents "Trajectory Clustering Based on Multi View Similarity" technique for clustering. The authors have demonstrated the efficiency of the proposed technique by developing Java based tool, called TCMVS and have experimented on real datasets.