pattern-based moving object tracking

ISCAS OpenIR

	pattern-based moving object tracking
	Liu Kuien; Ding Zhiming; Li Mingshu; Deng Ke; Zhou Xiaofang
	2011
会议名称	2011 International Workshop on Trajectory Data Mining and Analysis, TDMA'11, Co-located with UbiComp 2011
会议录名称	TDMA'11 - Proceedings of the 2011 International Workshop on Trajectory Data Mining and Analysis
页码	5-14
会议日期	September
会议地点	Beijing, China
收录类别	EI
ISBN	9781450309332
部门归属	(1) Institute of Software Chinese Academy of Sciences Beijing China; (2) School of ITEE University of Queensland Brisbane Australia
摘要	Monitoring the locations of a large number of objects that travel in a certain space is a popular problem for its importance in various application scenarios. It brings us a challenge of how to efficiently handle large volumes of location updates required to guarantee the error bound among an object's current, actual location and its current location in the tracking system. Current solutions predict the future locations based on the recent movements of the moving object. However, it is reliable to predict the position in near future only and the prediction accuracy is poor in the long term. This paper is aimed at the above weakness by introducing the movement pattern in Euclidean space based on the historical trajectories of moving objects. Dominant path pattern is proposed and employed in the moving object tracking system, which can estimate where an object will go next and how to get there. Specifically, dominant path pattern is discovered and indexed by a novel access method of efficient query processing. In addition, the pattern mining techniques with consideration of the accuracy and coverage in dominant path patterns discovering are presented. The experiments demonstrate the superiority of the proposed method comparing to existing methods by up to 73%(91%) less overall location updates on practical Taxi(Truck) dataset. Copyright 2011 ACM.; Monitoring the locations of a large number of objects that travel in a certain space is a popular problem for its importance in various application scenarios. It brings us a challenge of how to efficiently handle large volumes of location updates required to guarantee the error bound among an object's current, actual location and its current location in the tracking system. Current solutions predict the future locations based on the recent movements of the moving object. However, it is reliable to predict the position in near future only and the prediction accuracy is poor in the long term. This paper is aimed at the above weakness by introducing the movement pattern in Euclidean space based on the historical trajectories of moving objects. Dominant path pattern is proposed and employed in the moving object tracking system, which can estimate where an object will go next and how to get there. Specifically, dominant path pattern is discovered and indexed by a novel access method of efficient query processing. In addition, the pattern mining techniques with consideration of the accuracy and coverage in dominant path patterns discovering are presented. The experiments demonstrate the superiority of the proposed method comparing to existing methods by up to 73%(91%) less overall location updates on practical Taxi(Truck) dataset. Copyright 2011 ACM.
关键词	Data Mining Forecasting Navigation Taxicabs Trajectories
主办者	ACM SIGCHI; ACM SIGMOBILE
语种	英语
内容类型	会议论文
URI标识	http://ir.iscas.ac.cn/handle/311060/16210
专题	中国科学院软件研究所
推荐引用方式 GB/T 7714	Liu Kuien,Ding Zhiming,Li Mingshu,et al. pattern-based moving object tracking[C],2011:5-14.