Using VR-tree Index to Fast Find Trajectory Stop Regions

Discovering stop regions is an important topic of trajectory data analysis.A novel method using VR-tree index structure is proposed for fast extracting trajectory stops.Firstly,one trajectory is selected and whose stops are found using DBSCAN algorithm,then VR-tree is built according to these stops which are considered as candidate stops of other trajectories.Then,for rest trajectory,compare with the stops storied in VR-tree to find intersecting parts,mark these points which are contained in these regions.Other points of the trajectory are used to find new stops through DBSCAN algorithm,which will be inserted into VR-tree as the candidate of other trajectories.Stops of trajectory are obtained by merging the stops which are composed of two parts,one comes from the results of comparing with VR-tree,and another is the new stops,according to the coverage with each other.Experiments using real data show that comparing with the density based methods,this method not only can find stops very well,but increase efficient nearly 60%.