Design of Human Tracking Algorithm Based on Improved Camshift

Considering the problem of added background interference when initial target region is chose in Camshift tracking algorithm. This paper proposes a human tracking algorithm based on improved Camshift. The algorithm uses weight to determine the type of pixels in the back-projection, and then convert the back-projection into binary image, so as to improve the input accuracy of Camshift processing function. According to the human body size information, the algorithm appropriately improves the ratio of major axis and minor axis of search window, to optimize the output size of Camshift processing function. The contrast experiment proves the algorithm obviously improve accuracy and robustness of tracking effect.