People tracking and abnormal situation detection in the open space

Recently, social security problem in the public place is more important, ex. stealing, losing and some dangerous action, and how to detect and prevent them on demand now is an important task. In this paper, we present a method of tracking objects and detecting abnormal situation in the open space. Usually, background model determines a robust foreground region. Therefore, we first propose a new algorithm to estimate background. Then, a new tracking people algorithm is proposed. It not only can track people, but also can recognize people from merge to split. Final, we detect the abnormal situation in the open space. The method is computationally fast enough to track people in real time.