为了提高基于指尖的人机交互系统的环境适应能力和指尖检测的精准度,针对肤色易受环境中类肤色区域的干扰和指尖特征不明显的问题,在研究YCbCr色彩空间各参量与肤色的关系基础上,提出将(Cr+Cb)/2与(Cr-Cb)作为新的肤色识别参量进行手部提取,利用子区域对手部轮廓进行优化,突出指尖特征,并依靠所建立的轮廓点状态描述方程对指尖进行检测和跟踪。实验结果表明该方法能有效排除类肤色区域和手指轮廓不足对指尖检测造成的影响,具有较高的检测精度。

(Cr+Cb)/2 and(Cr-Cb)as new skin color identification parameters based onYCbCr skin color model are presented in order to improve the accuracy of fingertipsr detection and to enhance the applications of Human Computer Interaction (HCI) based on fingertips, dealing the interference of skin-like regions on complex background and the unobvious fingertips' features on hand contour. The hand contour is improved by subareas to be more prominent, which is detected and tracked by the subareas. Experimental results indicate that this method has better effect on ruling out skin-like regions and optimizing the fingertip features, and it has high accuracy on fingertips detection.