Algorithm of object hovering detection and tracking based on moving track

This paper provides an algorithm of irregular behavior recognition which is based on moving track. The algorithm can track hovering objects efficiently in the protection area and be competent for unmanned intelligent monitoring. Firstly, background subtraction method is used to detect objects. Secondly, the combined algorithm with modified Camshift (continuously adaptive mean shift) and Kalman filter is utilized to track objects. According to the moving track, we determine whether objects are suspicious or not. Finally, experiment shows that the algorithm has good reliability, robustness and also meets the real-time requirement.