Abstract: This paper presents an approach of deformable hand gesture tracking with single camera under disturbances from skin-color.It proposes a method to recognize static hand gestures based on PGH.In order to solve the problem of disturbances from skin-color，it achieves a predictive tracking method based on Kalman filter.Moreover，it proposes a solution based on a hierarchical structure to solve the problem of tracking initialization.Experimental results show that the proposed method can track the deformable hand gesture under disturbances from skin-color effectively，and thus it can meet the requirements of real-time vision-based human-computer interaction.

摘要： 介绍了一种基于单目视觉的肤色干扰下的变形手势跟踪方法。根据跟踪过程中所用到的基本手势特征，提出了一种基于PGH（成对几何直方图）的静态手势识别方法。为了解决跟踪过程中的肤色干扰问题，实现了基于Kalman滤波器的手势预测跟踪。为了解决跟踪过程中的初始化问题，提出了一种基于层次结构的跟踪初始化解决方案。实验结果表明，该方法能够在肤色干扰的情况下有效地对变形手势进行跟踪，并能够满足基于视觉的实时人机交互的要求。