基于计算机视觉的3D手势交互系统

3D gesture interactive system based on computer vision

随着计算机的广泛发展,键盘、鼠标等传统的人机交互方式很难满足用户自然、便捷的交互需求。研究手势建模、人手跟踪和手势交互系统的应用成为热点趋势。提出了一种简化的2D人手模型,该模型将人手建模为掌心点和5根手指,同时设计了一种基于粒子群优化(PSO)算法的人手跟踪方法,通过建模人手的生理和运动学约束关系,实现了基于2D/3D人手模型的PSO人手跟踪,该手势交互系统框架更具适用性和扩展性,融合了语义和反馈信息,提高了人手跟踪的鲁棒性和手势识别的准确度。

With the wide development of the computer,the keyboard,mouse and other traditional human-computer interaction modes are difficult to meet the users′ natural and convenient interaction needs. The study on the applications of gesture modeling,hand tracking and gesture interactive system has become a hotspot. A simplified 2D hand model is proposed in this paper,in which the hand is modeled as the palm point and 5 fingers. A hand tracking method based on particle swarm optimization（PSO）algorithm was designed,and the PSO hand tracking based on 2D/3D hand model was realized by modeling the physiology and kinematics constraint relation of the hand. The framework of this gesture interactive system has applicability and scalability. It fused the semantics and feedback information,and improved the robustness of hand tracking and accuracy of gesture recognition.