摘要
利用微软Kinect体感设备进行骨骼数据采集,开发出一款利用手势动作控制计算机的体感虚拟鼠标软件。并在开发过程中设计并提出一种基于RGB-D信息的人体手势动作检测及识别方法来处理Kinect输出的彩色影像和深度影像。该方法分别利用DBSCAN与K-means聚类算法获取手势操作特征中的位置信息和方向信息来识别手势操作,实验结果证明了该方法的可行性。借助基于该方法开发的虚拟鼠标软件,用户只需要做出一些简单的手臂动作即可操作虚拟鼠标完成对计算机的控制。
The author develops virtual mouse software which can control computers by users' gestures with the help of the skeleton data collected by Microsoft Kinect.This article also designs a novel method which can detect and recog-nize users' gestures to process the RGB stream and Depth stream exported by Kinect.This novel method leverages the DBSCAN and K-means clustering algorithms respectively to catch the location and direction information among the gesture features.The experimental results show the feasibility of this novel method.With the help of this virtual mouse software based on this method,users can simply make some gestures control the computers.
出处
《软件》
2016年第2期46-49,共4页
Software
