A real time approach to track humans in surveillance videos

Video surveillance is one of the recent research topics in computer vision. The goal of video surveillance is to gather information from videos by tracking the people involved in those videos and understanding the behavior of them. In general, Object tracking is an essential block in video surveillance system to identify and track objects present in it. After object identification, estimation of trajectories of the objects are crucial to understand and track its motion. In this paper, object tracking has been done through particle filter with likelihood function. Initially, Background has been modeled with the mixture of Gaussians using GMM. Later, using particle filters, trajectory has been drawn. This paper attempts to track a particular person in static and dynamic environments also. Results obtained through this tracking seems to be promising.