手势识别系统中手指及指尖检测方法

Finger and fingertip detection for hand gesture recognition system

为了满足手势识别系统中手势的识别与理解的需要,提出了一种快速、鲁棒的手指及指尖检测方法。首先根据肤色在颜色空间中的聚类特性,初步分离出感兴趣的手势部分。然后在此基础上,通过手轮廓边缘曲率特征法结合用于检测平行线的环形算子,实现指尖和手指的快速精确检测。实验结果表明,该方法能有效的检测出指尖、手指,并为进一步在手势识别系统中应用打下基础。

In order to meet the demand of hand gesture recognition system,a fast robust hand segmentation and fingertip detection method is proposed in the present article. First, the gesture part of interest is separated primarily according to the clustering property of skin color space. And then,by applying contour curvature method in combination with ring op- erator used for detecting parallel line, high-level feature of the gesture,i, e. , fingertip and finger,is extracted. Experiments show that the method is capable of effectively detecting fingertip and finger,and is capable of being applied in the gesture recognition system. And lay the foundation for further application in hand gesture recognition system.