您的位置:首页  > 论文页面

GPS轨迹中活动停留点识别的多层分割算法

发表时间:2011-04-30  浏览量:3292  下载量:1445
全部作者: 张治华,季民河
作者单位: 华东师范大学地理信息科学教育部重点实验室
摘 要: 个人移动通讯和位置感知设备的广泛使用产生了大量可用于信息服务的出行轨迹数据。从轨迹数据挖掘出行信息的关键在于停留识别和语义标注。已有的停留点识别方法在抗噪能力和计算效率上有所不足,识别精度有待提高。重新分析了轨迹的停留和移动两大组成要素,发现其状态存在的基础在于其在时间或空间上的连续性,并基于这一理念提出了一种多层次分割算法实现对轨迹停留点的识别。方法的实证检验使用了GPS模块收集的上海市11位受访者一周的出行活动轨迹及问卷调查表。实验结果表明:多层分割法在精度和计算效率上均显示出较好效果。
关 键 词: 模式识别;GPS语义轨迹;多层分割;活动停留
Title: Hierarchical segmentation for identifying activity stops from GPS trajectories
Author: ZHANG Zhihua, JI Minhe
Organization: Key Lab oratory of Geographic Information Science, China’s Ministry of Education, East China Normal University
Abstract: In this paper, currently available methods for identifying activity stops from GPS travel trajectories were examined, and their inability to accommodate data noise and lack of computational efficiency were identified. Based on the observation of multi-level activities imbedded within the trajectory, a hierarchical segmentation method was proposed to cope with the issue of varying spatial scales in which different activities take place. By recognizing the interweaving relationship between stops and moves, the new method starts out from generating basic segments from adjacent GPS points, ten combines adjacent segments into state segments according to their duration similarity, and further brings in some other attributes to determine activity stops and trips at different spatial scales. The algorithm was tested with sample GPS track data collected from 11 survey respondents over a week in Shanghai. Results indicated that the multi-level segmentation method could improve both classification accuracy and computational efficiency over the traditional density-based segmentation methods.
Key words: matrix recognition; semantic GPS trajectories; hierarchical segmentation; activity stops
发表期数: 2011年4月第8期
引用格式: 张治华,季民河. GPS轨迹中活动停留点识别的多层分割算法[J]. 中国科技论文在线精品论文,2011,4(8):673-682.
 
0 评论数 0
暂无评论
友情链接