Robust initial background extraction algorithm based on dynamic analysis

Traditional surveillance system needs a lot of manpower to search video for finding the invaders. Therefore, using manpower to search the videos is very unprofitable. Hence, the developments of intelligent surveillance system are very important issue for surveillance system. object segmentation method is most important technique on intelligent surveillance system, because the results of object segmentation can support intelligent surveillance system to detect and monitor Intrusive object. And background subtraction method is very popular of object segmentation. However, background subtraction methods have a problem in initial background extraction when the scene exist slow-speed objects. Because the discrimination between slow-speed objects and background are very difficult. In the paper, we propose a robust initial background extraction algorithm applied to video sequences with a static camera. The algorithm can classify and extract the background effectively. Even in slow-speed objects in the video sequence, the proposed algorithm can extract the initial background by dynamic analysis and maintained the background accurately.