Multi-objects tracking research of human body motion based on feature center

It is widely used to track the moving human body by Kalman filtering technology in intelligent video surveillance system. But in fact multi-human bodies´ motions often appear and when one is sheltered by another the object is lost. This paper mainly discusses this problem and solves it by means of extracting the features of moving human bodies and utilizing the technology of self-adaptive Kalman filtering. It takes the human feature centre prediction as an aided method to achieve the continuous tracking to moving multi-human bodies and avoid the object´s losing while two moving human bodies separate from the fixed objects. It improves the effectiveness and robustness in tracking system.