Abstract

Coherent motions, which describe the collective movements of indi- viduals in crowd, widely exist in physical and biological systems. Understand- ing their underlying priors and detecting various coherent motion patterns from background clutters have both scienti fic values and a wide range of practical ap- plications, especially for crowd motion analysis. In this paper, we propose and study a prior of coherent motion called Coherent Neighbor Invariance, which characterizes the local spatiotemporal relationships of individuals in coherent mo- tion. Based on the coherent neighbor invariance, a general technique of detecting coherent motion patterns from noisy time-series data called Coherent Filtering is proposed. It can be effectively applied to data with different distributions at different scales in various real-world problems, where the environments could be sparse or extremely crowded with heavy noise. Experimental evaluation and comparison on synthetic and real data show the existence of Coherence Neighbor Invariance and the effectiveness of our Coherent Filtering.1