Multi-feature Fusion Target Tracking Algorithm Based on Global and Local Consistency

Complex environment has a great impact on robustness of target tracking, particle filter based on normalized DPM histogram and meanshift tracking algorithm based on local HSV+LBP fusion characteristics was combined together in this paper, and the global and local consistency was constructed to complete target tracking. The algorithm combines robust global color information with local texture information, and a global-local estimation model with adjustable weights was constructed, and the current state of the target was obtained. The experimental results show that the algorithm can deal with occlusion, interference, fast moving of the target, and finally achieve the stable tracking of the target.