基于TLD的目标追踪算法

s an important research direction of computer vision, visual tracking becomes attractive to more and more researchers. Currently, most of the studies focus on the scale of the target or tracking accuracy, without considering the problem such as the target disappeares or be completely obscured. This paper proposes a TLD tracking algorithm based on Mean Shift. This method not only focuses on the tracking algorithm, but also combines the tracking algorithm and detection algorithm. With the help of the learning algorithm, the tracking block and detection block can correct the error of each other. The experiments show the algorithm could effectively reduce tracking failure and get better tracking performance compared with the original TLD algorithm.