Object Tracking Algorithm Based on Meanshift Algorithm Combining with Motion Vector Analysis
Gang Tian, Ruimin Hu, Zhongyuan Wang, Li Zhu · 2009
Mean shift algorithm doesn't use the targetpsilas motion direction and speed information in process of object tracking. When the targetpsilas speed is so fast it easily fails to track the target. So a new object tracking algorithm combining Mean shift algorithm with Motion Vector analysis is proposed in this paper. By statistical analysis of the motion vector get from video encoding process, we can get the motion direction and velocity of target, which can be used to correct the central point of the motion candidate region of Mean shift, making the search position is more close to the actual centre of the target. This method can not only track the fast moving target effectively, but also reduce the number of iterative convergence times to improve the efficiency of operations. The algorithm is already use in our intelligent video surveillance equipment in which the operation of video encoding and object tracking is executed in one chip, and the experimental results show that it is feasible and effective.