Study On Spatiotemporal Data Based Moving Object Similarity Measurement

With the popularity of mobile devices and the development of positioning technology,the spatialtemporal data generated by the moving object is explosive growth,how to dig out valuable information from the massive data has received more and more attention.Moving objects similarity measurement has broad application prospects in the fields of trajectory data mining and location-based services.This thesis carries on the thorough study to moving objects similarity.Firstly,according to the characteristics of the moving object activity,a trajectory structure based hotspot discovery framework is proposed,which solves the problem that the existing algorithms only consider the trajectory density characteristic.On the basis of this framework,a candidate hotspot discovery algorithm and a hotspot filter algorithm are given.Candidate hotspot discovery algorithm uses relative density clustering to solve the problem of hotspot discovery in multi-density spatiotemporal data,and can effectively identify potential activity-intensive areas.The hotspot filter algorithm fully examines the activity characteristics of trajectory in candidate region,and identifies the active hotspot according to the structure attributes and time attributes of trajectory.Then,a frequent activity pattern based moving object similarity measurement is given.Most hotspot discovery algorithms view trajectory similarity as moving object similarity,which cause the algorithm can not compare the activity habit of moving objects.Based on the found activity hotspots,the spatiotemporal trajectories are transformed into active sequences,which carry more semantic information.Frequent activity patterns representing the motion of the moving object can be excavated from the active sequence.The similarity of the moving object is calculated by the frequent activity patterns.Finally,a prototype system based on the theoretical research for similar moving object query is designed and implemented to verify the method proposed in this paper.The effectiveness of the method is verified by the visualization.We take a modular development approach to make the system more scalable.