局部多层网格划分下的轨迹数据概化方法  

Trajectory data generalization based on local multi-hierarchy grid

作　　者：杨光[1] 张磊[1] 李帆[1] 

机构地区：[1]中国矿业大学计算机科学与技术学院,江苏徐州221116

出　　处：《计算机应用》2013年第6期1604-1607,共4页journal of Computer Applications

基　　金：教育部博士点基金资助项目(20110095110010);江苏省博士后基金资助项目(0802023C);中国矿业大学青年科技基金资助项目(2008A040);中国矿业大学研究生实践与科研创新课题专项基金资助项目(GSF122111)

摘　　要：针对轨迹数据概化中空间划分的区域范围不能有效控制以及覆盖网格尺度难以合理选择的问题,提出局部多层网格划分方法,对样本密集的区域进行迭代划分。在此基础上提出一种轨迹数据概化方法,在局部多层网格划分的基础上,考虑时间约束合并轨迹连续往复通过的邻接区域,生成概化轨迹。真实数据的实验表明该算法得到的概化轨迹较同类算法保持了更多轨迹特性,更加适合后续数据挖掘,如聚类处理。For current trajectory data generalization methods, the scope of the generalized regions cannot be controlled effectively, and the parameters of the grids earl hardly be selected logically. This paper proposed the method of Local Multi- hierarchy Grid （LMG）, so that the region with dense trajectory points would be divided iteratively. And then a method for trajectory data generalization named TRAGenLMG was proposed, which was based on LMG, and time-constraint was used to merge some adjoining grids, and finally the generalized trajectory was got. The experiments with real open dataset show that the generalized trajectories generated by TRAGenLMG can well maintain the temporal feature of the trajectory data and can be efficiently applied into further data analysis.

关 键 词：轨迹概化 局部多层网格 时间约束 轨迹数据 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]