摘要
基于小样本库的手势识别是先进人机交互研究中的一个重要分支.根据Tortoise人手模型训练手势模式库并结合交互者的具体手部特征进行手形训练,生成适用于特定交互者的手势模式库.在交互过程中,根据来自一个或多个同步摄像头的视频信息进行基于自适应遗传算法的手势识别.实验结果表明,在环境光照基本稳定的条件下,文中算法可以实现鲁棒的实时手势识别.
Gesture recognition based on small samples is one of the main trends in the advanced humancomputer interaction research. This paper tries to develop a general hand-shape model database based on the Tortoise Model （TM）. Before the interactions, we use the user＇s hands information to update the general gesture model database and create the user-specified gesture database. Thereafter, during the course of interactions, via the video information from one or more synchronous cameras, we match the user＇s certain gesture with the user-specified gestures based on adaptive genetic algorithms. The experimental results show that the robustness of real-time gesture identification can be improved greatly if the ambient light is relatively stable.
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2007年第8期1056-1062,共7页
Journal of Computer-Aided Design & Computer Graphics
基金
国家自然科学基金(60021201)
国家"九七三"重点基础研究发展规划项目(2002CB312102)
关键词
虚拟现实
手势交互
人手模型
手势识别
自适应遗传算法
virtual reality
gesture-based interaction
human hand model
gesture identification
adaptive genetic algorithm
