三维人手关节点的提取研究*

近年来以计算机为载体的人机交互愈发成为人们日常生活中的一个重要组成部分,人手作为仅次于语言的交流方式,有着不可或缺的作用,在人机交互中作为主要媒介有着重要的研究价值。本文在关节点的提取上基于特征匹配提出了一种新的研究方法-跨维匹配。在立体视觉手势交互的基础上,用kinect相机采集带有关节点标记的彩色图像和三维点云,通过特征点提取和匹配的方法,获得手势点云平面图的屏幕坐标,并与三维手势点云一一对应得到手势关节点的三维坐标。实验表明本文的方法是可行的,并且在手势关节点的提取效果上也很精确,能够很容易的应用到手势变换中。

#$TABThe research of extracting on the 3D joints from human hands#$TAB

In recent years, the computer as the carrier of human computer interaction has become an important part in People's Daily life, hand as a second language of communication has the indispensable role, as the main medium in the human-computer interaction has important research value. In this paper, a kind of new research methodsis proposed based on the joint point of the extraction of feature matching . The Kinect was used to collect the same gesture with joint point mark color images and 3d point cloud based on stereo vision gesture interaction, and through the methods of extracting and matching feature points, obtained gestures screen coordinates of the point cloud plan, and compared with 3d gesture point cloud by one-to-one correspondence to get key points of 3 d coordinate. Experiments show that the method is feasible, and the key points in the gestures of extraction effect is also very accurate and.can be easily applied to the tansform of the gestures.