Cam-shift target tracking approach based on back projection method with distance weights
Wei Gang Hu, Yuqiu Sun, Xiong Li, Yuanxiang Jiang, Meichen Yu · 2016
Traditional Cam-Shift Target Tracking approaches are easily affected by complex environment. They maybe cannot distinguish the target from the environment and make a mistake recognition. A Cam-Shift Target Tracking approach with distance weights on the back projection is proposed in this paper. First, extract H component from target image based on the control factor N and calculate the histogram Hi of target region. Second, calculate the color probability (back projection) of next frame and weight the back projection by distance weights. Then, calculate barycenter and move the center of the searching window in second frame use Mean-Shift algorithm. It is shown from the experiment that the method can distinguish the target in a complex environment better.