摘要
基于单目视觉下的手势识别技术一般由手势建模、特征提取、手势匹配等几个关键技术构成。手势跟踪算法目前主流的是粒子滤波算法和Camshift算法。系统采用Camshift算法,将人手图像由RGB空间转换到HSV空间后,在HSV空间利用半自动预定义模板颜色对人手进行分割,并对其进行改进实现多目标跟踪,由于Camshift算法为半自动算法,在对手势进行跟踪前需对手势进行手动标定,系统采用了手势跟踪与手势识别技术结合的方法,改进了Camshift算法,解决了Camshift的半自动问题和实现多目标跟踪,实现双手的手势识别。
The gesture recognition technology based on monocular vision is usually constituted by the key technologies of gesture modeling,gesture segmentation,feature extraction and gesture matching.The most popular gesture tracking algorithms based on monocular vision are the particle filtering algorithm and Camshift algorithm.This system converts the hand images from RGB space to HSV space,segments the hands using templates predefined color value,makes some improvements and realizes multi-objective target tracking.Because Camshift algorithm is semi-automatic,the gesture must be calibrated manually before being tracked.The system combines the gesture tracking and gesture recognition technology,improves the Camshift algorithm,solves the problem of being semi-automatic,and realizes multi-objective target tracking and finally realizes the gesture recognition of both hands.
出处
《电子科技》
2012年第2期71-73,81,共4页
Electronic Science and Technology
