Incremental clustering for trajectories

Trajectory clustering has played a crucial role in data analysis since it reveals underlying trends of moving objects. Due to their sequential nature, trajectory data are often received incrementally, e.g., continuous new points reported by GPS system. However, since existing trajectory clustering algorithms are developed for static datasets, they are not suitable for incremental clustering with the following two requirements. First, clustering should be processed efficiently since it can be frequently requested. Second, huge amounts of trajectory data must be accommodated, as they will accumulate constantly. An incremental clustering framework for trajectories is proposed in this paper. It contains two parts: online micro-cluster maintenance and offline macro-cluster creation. For online part, when a new bunch of trajectories arrives, each trajectory is simplified into a set of directed line segments in order to find clusters of trajectory subparts. Micro-clusters are used to store compact summaries of similar trajectory line segments, which take much smaller space than raw trajectories. When new data are added, micro-clusters are updated incrementally to reflect the changes. For offline part, when a user requests to see current clustering result, macro-clustering is performed on the set of micro-clusters rather than on all trajectories over the whole time span. Since the number of micro-clusters is smaller than that of original trajectories, macro-clusters are generated efficiently to show clustering result of trajectories. Experimental results on both synthetic and real data sets show that our framework achieves high efficiency as well as high clustering quality.