基于Mean Shift和灰色预测的快速运动行人跟踪

针对MeanShift跟踪算法在跟踪快速运动的行人时往往不能持续跟踪的不足，提出了一种基于GM（1，1）灰色预测的MeanShift运动行人跟踪算法。依据行人衣装颜色分布建立了目标模板，利用GM（1，1）模型预测行人可能出现的位置并作为搜索目标的起始点，在该点邻域内利用MeanShift迭代求取行人的实际位置。实验结果表明，本算法对于快速运动行人和被遮挡行人均能给出较为良好的跟踪效果，并减少了MeanShift的迭代次数。

Fast-moving Pedestrian Tracking Based on Mean Shift and Gray Prediction

To address the issue that the Mean Shift algorithm can not keep tracking pedestrians when pedestrians move fast, this paper presents a new pedestrian tracking algorithm using Mean Shift with GM （1, 1 ） gray prediction model. The algorithm establishes the target template based on the clothes color of the pedestrian, uses GM （1, 1 ） model to predict the approximate pedestrian location which is used as the starting point for search, and calculates the real location of the pedestrian by Mean Shift iterative algorithm in the neighborhood of the point. Experimental results show that the proposed algorithm can give good track performance whether pedestrians moving fast or blocking each other, and simultaneously reduce the number of iterations.