Research On Trajectory Tracking Based On Spatio-temporal Data Mining

With the rapid development of information technology,mobile terminals with positioning functions are used in various fields,causing a large amount of spatio-temporal trajectory data to be continuously generated.A reasonable mining method can dig out the hidden information behind it while reducing data redundancy.Based on the research and analysis of data mining algorithms and spatiotemporal frequent trajectory pattern mining algorithms,this paper proposes a hotspot region discovery algorithm and a region reconstruction algorithm based on spatiotemporal trajectory data.At the same time,the algorithm of frequent trajectory pattern mining in time and space is researched and improved.The main research contents are as follows:First of all,in order to solve the problem that the existing spatial clustering methods are expensive and not suitable for the characteristics of spatio-temporal data,an algorithm DP_Cluster based on grid clustering and spatiotemporal density clustering is proposed.Grid division and trajectory projection are carried out in the space,and the number of density direct points existing in the neighborhood of each point and the spatial distance between points are calculated,and the core trajectory points that meet the density threshold and the neighborhood are found Combine,get the candidate hotspot area of spatiotemporal data.Secondly,in order to solve the problem that the density clustering algorithm classification effect is not ideal due to the uneven data density of spatiotemporal trajectory data,a hotspot area reconstruction algorithm based on area and trajectory spatiotemporal attributes is proposed,DP_Rebuild.This method can reconstruct the clusters that do not meet the area constraint,solve the problem that the cluster area is too large and the hotspot area features are not prominent,and improve the classification effect of the density clustering algorithm on uneven density data.Finally,in order to more effectively mine the frequent patterns of spatio-temporal sequence data,a frequent pattern mining algorithm based on spatio-temporal trajectory,BOTM-mining,is proposed to find frequent position sequences in spatio-temporal trajectories.According to the proposed frequent pattern processing model,the spatiotemporal trajectory in geographic space is first converted into a trajectory sequence with time attributes,and then the support information,sequence information,and mining location in the spatiotemporal data are separated by constructing a processing model to compress and store the spatiotemporal information.The required index information is then used to recursively generate a projection database based on the prefix tree method to complete the mining.The experimental results show that the method is effective and feasible.