地理科学进展 ›› 2011, Vol. 30 ›› Issue (5): 522-534.doi: 10.11820/dlkxjz.2011.05.002

• 研究综述 • 上一篇    下一篇

时空轨迹聚类方法研究进展

龚玺1,2, 裴韬1, 孙嘉2,3, 罗明4   

  1. 1. 中国科学院地理科学与资源研究所资源与环境信息系统国家重点实验室, 北京 100101;
    2. 中国科学院烟台海岸带研究所, 烟台 264003;
    3. 中国科学院研究生院, 北京 100049;
    4. 香港中文大学地理与资源管理学系, 香港
  • 收稿日期:2010-10-01 修回日期:2011-02-01 出版日期:2011-05-25 发布日期:2011-05-25
  • 通讯作者: 裴韬(1972-),男,副研究员,主要从事空间数据挖掘和空间信息统计等的研究。E-mail: peit@lreis.ac.cn E-mail:peit@lreis.ac.cn
  • 作者简介:龚玺(1986-),男,硕士研究生,主要研究方向为空间数据挖掘。E-mail: gongx@lreis.ac.cn
  • 基金资助:

    中国科学院知识创新工程重要方向项目(KZCX2-YW-QN303);中国科学院地理资源所自主部署创新项目(200905004);863 项目(2009AA12Z227)。

Review of the Research Progresses in Trajectory Clustering Methods

GONG Xi1,2, PEI Tao1, SUN Jia2,3, LUO Ming4   

  1. 1. State Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China;
    2. Yantai Institute of Coastal Zone Research, CAS, Yantai 264003, Shandong,China;
    3. Graduate University of Chinese Academy of Sciences, Beijing 100049, China;
    4. Department of Geography and Resource Management, The Chinese University of Hong Kong, Hong Kong, China
  • Received:2010-10-01 Revised:2011-02-01 Online:2011-05-25 Published:2011-05-25

摘要: 时空轨迹(Trajectory)是移动对象的位置和时间的记录序列。作为一种重要的时空对象数据类型和信息源,时空轨迹的应用范围涵盖了人类行为、交通物流、应急疏散管理、动物习性和市场营销等诸多方面。通过对各种时空轨迹数据进行聚类分析,可以提取时空轨迹数据中的相似性与异常特征,并有助于发现其中有意义的模式。本文根据时空轨迹数据的特点,系统综述了时空轨迹聚类方法的研究进展。首先,从理论、可行性和应用的角度分析了时空轨迹数据及其聚类方法研究的重要性,并论述了时空轨迹的定义、模型与表达;然后,按照相似性度量所涉及的不同时间区间将现有的时空轨迹聚类方法划分为6 类,并对每一类方法的原理及特点进行了评述;最后,讨论了现有方法面临的主要问题和挑战,并对时空轨迹聚类研究的发展进行了展望。

关键词: 聚类, 时空轨迹, 时空数据挖掘, 相似性度量, 研究进展

Abstract: A trajectory is a sequence of the location and timestamp of a moving object. It is not only an important type of spatio-temporal data, but also a critical source of information. Extracting patterns from different trajectory data can help people understand the drives and outcomes of individual and collective spatial dynamics, such as human behavior patterns, transport and logistics, emergency evacuation management, animal behavior, and marketing. Recently, a larger number of trajectory data are available for analyzing the temporal and spatial pattern, as the result of the improvements of tracking facilities and sensor networks. Therefore, clustering analysis needs to be used to find the implicit patterns in it. Based on the characteristics and the similarity measurements of trajectory data, this paper reviewed the research progresses in trajectory clustering methods. Firstly, the significance of research on trajectory data and its clustering methods was presented. Then the definition, models as well as several visualization methods of trajectories were summarized. After that, the authors classified the existing trajectory clustering methods into 6 main categories according to the similarity measurement of them, and analyzed each of the trajectory clustering methods, along with their respective pros and cons by category. Finally, some research challenges and future directions were discussed.

Key words: clustering, research progress, similarity measurement, spatio-temporal data mining, trajectory