鉴于Mean Shift算法中核心步骤计算所得的新目标模板中心位置几乎都非整数的问题,该文提出了模糊隶属度的概念,并在此基础上,改进了Mean Shift算法在图像跟踪中的实现步骤。实验表明,该算法计算量小,且能很好的减少原算法中不必要的误差。
In view of the fact that the center position of new target template obtained by the computation of the core step of Mean Shift algorithm is nearly all non-integer, in this paper, a concept of fuzzy membership is proposed, the steps in image tracking using Mean Shift algorithm are improved. The results of experiments show that computational load and error are less than that of the original algorithm.