Mixture of Gaussian Based Background Modelling for Crowd Tracking Using Multiple Cameras

Visual surveillance system for tracking crowd using multiple cameras at dynamic backgrounds faces many challenges such as illumination variance, occultation, low spatial temporal resolution, sleeping person, shadows and camera noise. In this paper we address the issue of gradual and sudden illumination variance caused by movement of the sun and the clouds. We evaluate Mixture of Gaussian method and background modelling method for extracting foreground from the background for crowd related data base. We have evaluated the performance of the background model for sparse and dense crowds to evaluate the accuracy and efficiency of the model subjectively for crowd analytics based scenarios.