一种快速鲁棒的动态人手跟踪方法

Fast and robust approach to tracking moving hand

介绍了一种基于肤色与手形的动态人手跟踪方法。根据检测过程中所用到的人手基本特征,提出了一种基于色度-饱和度自适应阈值肤色分割与成对几何直方图(PGH)匹配的人手图像检测方法。为了克服跟踪过程中背景色干扰的问题,实现了基于Kalman滤波器的预测跟踪。实验结果表明,该动态人手跟踪方法计算简便、实时性好、跟踪精度高,可应用在大多数复杂场景。

This paper presents an approach of moving hand tracking based on skin-color and hand shape. It proposes a method to detect moving hand based on chromatic adaptive thresholding and PGH matching. In order to solve the problem of the disturbances from background color, it achieves a predictive tracking method based on Kalman filter. The results show that the method is simple and accurate, and can be used in most complex scenes.