Background Modeling and Subtraction Based People Counting for Real Time Video Surveillance

This paper presents a multiple people counting using only single camera, entering or exiting in a region of interest. Sigma-Delta background modeling and subtraction is used to segment the people from region. Background subtraction provides blobs as a result with connected component. CG of blobs was used to track and count the entry and exit of people. This paper also proposed method for counting two or more people based on blob size and distance between floor and camera. People will be counted if and only if they passed through the three regions consecutively. Experiments shows that proposed method is robust and provides approximate accurate counts.