Sketch-based uncertain trajectories clustering
Jingyu Chen, Qiuyan Huo, Ping Chen, XU Xue-zhou · 2012
Uncertain trajectories data present new challenges to trajectories data mining. This paper proposes a sketch-based trajectory clustering algorithm for uncertain trajectories. Based on the M-level Hilbert curve spatial partitioning, a candidate segments set is constructed to represent uncertain trajectories model precisely. For the large number of candidate segments, a sketch-based approach is used to create hash-compressed clusters. A sketch-based clustering algorithm is proposed to assignment the incoming uncertain trajectory to clusters. The experiments prove that the clustering algorithm has stable accuracy with variations of the sampling rate of trajectories data.