基于BCH和HOG的Mean Shift跟踪算法

Mean Shift tracking algorithm based on BCH and HOG

为解决基于核的Mean Shift跟踪算法在颜色相近时,容易出现跟踪不准确或者跟踪丢失的问题,同时提高跟踪的可靠性与准确度,分析了近年来的目标跟踪情况,提出一种基于分块颜色直方图(BCH)和方向梯度直方图(HOG)的Mean Shift跟踪算法。其中BCH包含目标的空间信息,能够很好地应对形变;HOG能够很好地解决光照变化和阴影的影响。使用标准数据集进行测试,结果表明:所提算法可以达到高于80帧/s的跟踪速度,分别优于跟踪—学习—检测(TLD)和核相关滤波(KCF)算法1.7%和2.2%。

In order to solve the problem of that kernel-based meanshifl tracking algorithm is prone to occur, tracking inaccurate or tracking loss when the colors are similar, and at the same time, to improve tracking reliability and accuracy, after analyzing on situation of target tracking in recent years, a Mean Shift tracking algorithm based on block color histogram （ BCH ） and direction histogram of gradient （HOG） is proposed. Among them, BCH contains spatial information of the target and can well cope with deformation. HOG can well solve the influence of illumination changes and shadows. Use standard dataset for testing, the results show that the tracking speed can reach above 80 FPS, which is superior to 1.7 % and 2.2 % of TLD and KCF algorithm respectively.