摘要
提出了一种基于机器视觉的多点手势识别方法及其人机交互技术框架。指尖跟踪和手势识别服务程序通过一个普通的摄像机捕获用户手的运动,对多个指尖目标进行实时检测和跟踪,在指尖跟踪结果基础上利用BP神经网络实现多点手势识别,并根据指尖跟踪和手势识别结果构造相应的消息(包括低级指点消息和高级手势消息)发送给客户端应用程序,客户端响应消息并进行相应的处理。该框架可以帮助开发人员的在应用程序中增加类似iPhone多点触摸控制的多点手势识别控制功能,实现更加自然的人机交互,提高用户操作体验。
A framework of human-computer interaction based on multi-point gesture recognition is presented. The server of fingertip tracking and gesture recognition firstly captures the movement of user＇ s hands by a camera, detects and tracks multiple fingertips in real time, then realizes multi-point gesture recognition by making use of the results of fingertip tracking through BP neural network. Finally, the server constructs mcssages （including primary fingertip messages and senior gesture message） according to the result of fingertip tracking and gesture recognition and sends them the client application, the client responds to the messages. The framework can help programmers realize multi-point gesture based control functions just like multi-touch functions of iPhone, achieve a more natural human-computer interaction and improve the operation experience of user.
出处
《计算机工程与设计》
CSCD
北大核心
2011年第6期2129-2133,共5页
Computer Engineering and Design
基金
广东省科技计划基金项目(2009B030803031)
广东省自然基金项目(8152840301000009)
关键词
机器视觉
指尖跟踪
多点手势识别
人机交互
框架
machine vision
fingertip tracking
multi-point gesture recognition
human-computer interaction
framework
