Real time moving object detection for video surveillance based on improved GMM

At present automated surveillance system has become a new trend in the field of security and defence. The moving object detection is one of the fundamental steps for analysis of motion in video surveillance. It provides a way of classification of the pixels into the foreground and background. For a video processing system, the key step is to detect moving object and subtract the background. The mixture of Gaussian (MOG) models is the best suitable for systems having static and complicated background with clutters. It is efficient and robust to illumination changes and camera noise. It reduces the noise from the foreground of the image to a much lower level and causes an efficient detection of objects from a surveillance video that is helpful in many security operations and other applications such as people tracking, traffic monitoring etc. In this paper a new technique is presented to deal with the problem of slow moving objects and to provide fast object detection with robust noise removal and improved background updating. The experimental results show that the proposed method gives better results than the other traditional methods of background subtraction. The dataset that we are using here are CAVIAR an indoor sample video and one other standard outdoor video dataset.