摘要
很多经典的人脸识别方法难以适应姿势变化及人脸错位的情形。为了解决这一问题,提出一种基于纹理豪斯多夫距离(THD)的人脸识别算法。将人脸图像的空间量及纹理特征相结合,使其在深入的头部转动和人脸错位中都有很高的容错度。在FERET及Yale两大人脸数据库的实验表明,与其他经典的方法相比较,所提出的方法取得了更好的识别效果。
Many classical face recognition methods are hard to adapt to pose variation and face misalignment. To solve this problem, we propose a new face recognition method which is based on textural Hausdorff distance （THD）. The method combines the spatial amount of face image with textural features, this makes it have very high fault tolerance on both in-depth head turning and face misalignment. Experiments conducted on two big face databases of FERET and Yale demonstrate that the proposed approach achieves better recognition effect compared with other classical methods.
出处
《计算机应用与软件》
CSCD
北大核心
2014年第10期186-188,212,共4页
Computer Applications and Software
关键词
人脸识别
纹理豪斯多夫距离
子空间学习
判别分析
Face recognition
Texture Hausdorff distance
Subspace learning
Discriminate analysis
