The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Publications Copernicus
Download
Citation
Articles | Volume XL-2
https://doi.org/10.5194/isprsarchives-XL-2-55-2014
https://doi.org/10.5194/isprsarchives-XL-2-55-2014
11 Nov 2014
 | 11 Nov 2014

Trajectory Clustering for People's Movement Pattern Based on Crowd Souring Data

J. Chen, T. Hu, P. Zhang, and W. Shi

Keywords: GPS trajectories data; Spatial-temporal clustering algorithm; Movement pattern

Abstract. With the increasing availability of GPS-enabled devices, a huge amount of GPS trajectories recording people's location traces have been accumulated and shared freely on the Web. In this area, one of the most important research topics is to exploit trajectory-movement pattern about where and when people clustered based on the raw GPS data. In order to solve this problem, clustering is a good way to perform data mining tasks on trajectory data.

This paper provides a clustering algorithm which aims at mining people’s movement pattern about the clustered location and their temporal evolution characteristics. Firstly, the characteristic points of GPS trajectories were chosen. Based on the characteristic points, a trajectory has been partitioned into a group of line segments. These line segments can represent the movement pattern of trajectories much better than that of track points. Secondly, an improved density-based line clustering method was used for the individual partitioned line segments to find out individual clusters with similar track segments. In this step, the absolute time spot of people’s trajectories was taking into account as a characteristic for the temporal evolution of people’s trajectories. Finally, the representative clustered hot spots of multiple users’ line segments achieved by above steps were output. Experiments were conducted with GPS trajectories data downloaded from the web to verify the effectiveness of the algorithm in this paper. According to the results, the spatial distribution and temporal evolution characteristics of people’s stay hot spots were effectively discovered from people’s GPS trajectories data.