摘要
本文提出了一种人脸姿势估计方法,基于单幅人脸图像的五个特征点:左右外眼角,左右嘴角和鼻尖估计人脸姿势,论文首先给出人脸姿势表示——旋转角度((?),ψ,θ),然后介绍了姿势估计方法,最后的实验结果证明了本文算法的有效性。
in this paper, a method is proposed to estimate the facial pose from five feature points in a single face image : left outer comer of eye, right outer comer of eye, left comer of mouth, right comer of mouth and nose tip. First, two methods of facial pose representation are described, then the method of face pose estimation is described in detail. Finally, the experimental results confirm the efficiency of proposed method.
出处
《信号处理》
CSCD
北大核心
2006年第1期61-64,共4页
Journal of Signal Processing
基金
863项目(2003AA103960)北京工业大学博士启动基金(KZ0203200393)资助