基于正投影视图的多姿态人脸识别算法  

Multi-pose face recognition based on orthogonal views

作　　者：武芒[1] 王燕[1] 刘志镜[1] 李夏忠[1] 

机构地区：[1]西安电子科技大学计算机信息应用研究中心,陕西西安710071

出　　处：《西安电子科技大学学报》2004年第4期597-601,621,共6页Journal of Xidian University

摘　　要：提出了基于正投影视图的多姿态人脸识别技术,根据特定人的正投影视图建立出特定人的3D模型,将3D模型进行任意角度的投影产生出多姿态人脸图像,基于该正投影视图和生成的多姿态图像进行多姿态人脸识别.实验结果表明该算法识别的正确率远高于基于单前视图的算法.To deal with the problem of the traditional multi-pose face recognition based on the multi-view that it must have many face photos, a multi-pose face recognition algorithm based on the two orthogonal views is proposed. It consists of two steps. In the first step, a 3D face model is built by the person's two orthogonal views and multi-pose face images are produced by the projections of the 3D face model. In the second step, multi-pose face recognition is performed based on the training set that consists of the two orthogonal views and the synthesized multi-pose images. The experimental results show that the performance of the algorithm discussed in the paper is by far superior to that of the method based on the front view.

关 键 词：人脸识别 多姿态 正投影视图 3D模型 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]