手势数据挖掘及挖掘结果的可视化

Data Mining in Hand Gestures and Visual Presentation of the Result

摘要: 手势作为虚拟环境中的输入和控制装置,在人机接口中起着重要的作用.在对大量真实的运动手势数据进行聚类、得到所有的手势类的基础上,通过数据挖掘中的粗糙集方法对手势类进行约简,得到了一组基手势,任何手势都可以由这组基手势生成.该方法得到的结果自然、可信,挖掘的结果用VRML的三维虚拟人显示,较好地解决了模式的易懂性问题.

Abstract: This paper presents a data mining approach to cluster large amount of real hand gestures, then use rough set approach to eliminate redundant ones and come to 15 basic gestures which are able to compose any kind of real hand gestures.The obtained results are natural and reliable,they are presented visually by a virtual person with VRML,thus are easily understandable to everybody.