基于非参数聚类和多尺度图像的目标跟踪

提出了一种基于非参数聚类和多尺度图像的目标跟踪算法.在利用改进的非参数颜色聚类进行自适应划分目标颜色空间的基础上,定义了目标模型和候选目标模型,该模型利用高斯函数建模颜色直方图中的每一个颜色特征位的空域分布; 根据Bhattacharyya系数的定义得到了目标模型和候选目标模型之间的相似性函数.跟踪算法利用高斯金字塔得到的多尺度图像进行从粗到细的目标空间定位;同时通过利用推导的核函数自动带宽选择公式,实现了目标尺度定位.实验结果表明该方法优于典型的均值漂移跟踪方法,从而验证了该方法的有效性.

Object Tracking Based on Nonparametric Clustering and Multiscale Images

An object tracking algorithm based on nonparametric clustering and multiscale images is presented in this paper. Based on using the modified nonparametric clustering to adaptively partition the color space of a tracked object, the target model and the target candidate are defined. The spatial information of each bin of the color histogram is modeled as a Gaussian. The similarity metric between the target model and the target candidate is derived from the Bhattacharyya coefficient. The coarse to fine approach is employed to get the spatial location of the tracked object by using the multiscale images from the Gaussian pyramid. Besides, the derived automatic bandwidth selection of kernel function is used to obtain the scale of the tracked object. The experimental results show that the algorithm outperforms the classic mean shift tracker, which verifies the effectiveness of the algorithm.