Abstract
The purpose of moving targets tracking is to determine the trajectories of moving targets, the key of which is to establish the corresponding relationships between the detected prospect targets and the tracked moving targets. The established corresponding relations are equivalent to the matching problem of target features in consecutive image frames. In the paper, Kalman filter is used as the motion model of the targets. The recursive filtering method is adopted to calculate and predict the location of moving targets. Through the experiment, the results show that the algorithm not only track well the moving targets, but also can effectively overcome the obstacles in the process of tracking and so on.