基于视觉的多特征手势识别

手势是一种自然直观的交互方式,基于视觉的手势识别是实现新一代人机交互的关键技术。本文在已有的手势识别技术基础上,从手势分割及手势表示两方面着手,提出了一种单目视觉下的手势识别方法。利用颜色特征检测肤色区域,成功分割出人手;利用人手的轮廓及凸缺陷检测指尖,再利用指尖的数目和方位来表示一个手势,进而结合轮廓长度和面积等几何特征完成手势识别。传统的指尖检测方法需要遍历并扫描手掌外轮廓,计算量大,本文通过凸缺陷检测指尖,减少了计算量,提高了指尖检测的速度。实验结果表明,本文的方法具有很好的鲁棒性及实时性,能适应环境的变化。

Vision-Based Hand Gesture Recognition with Multiple Cues

Hand gesture is a kind of naturally intuitive interacting ways.Computer vision-based hand gesture recognition has become a key technique in the realization of new human-computer interaction.On the basis of the existing techniques of hand gesture recognition,this paper proposes a method in the recognition of hand gestures with a single camera from the facts of hand gesture segmentation and representation.The skin area is detected with color features and the hand is segmented successfully;the fingertips are detected with the hand’s contour and convex defects and then a hand gesture is represented with the number of fingertips and their positions,and therefore,the procedure of recognizing hand gestures is completed with geometric features such as the contour length and the area.The traditional approaches of detecting fingertips the process of retrieving and scanning the palm’s outer contour,resulting in very large computation.However,the paper reduces computation and improves the speed of fingertip detection by detecting the convex defect of fingertips.The experimental results show that the proposed method is robust and real-time,and is able to adapt to the environmental change.