DIVIDING AND CLUSTERING ALGORITHMS OF MOVING OBJECTS IN RFID TRACING SYSTEM
Kongfa Hu, Xiangqian Xue, Jianfei Ge, Yan Sun, Ling Chen · 2012
Trajectory clustering can predict moving trend of objects effectively. The traditional trajectory clustering algorithms take moving trajectory of a whole object as a research object, which will lose similar subtrajectories. However, in practical applications, such as in RFID system, the users may only focus on some specific regions of trajectories. We propose PT-CLUS algorithms in this paper, according to coarse-fine algorithm, which first dividing a trajectory into a group of line segments and prunes by coarse-fine strategy, and then searching cluster in the sub-trajectories by checking neighborhood region of segments, using hierarchical clustering to accomplish the clustering of sub-trajectories. Experiment result shows that PTCLUS algorithm can find the similar sub-trajectories from RFID trajectory database effectively.