用于人脸识别的GFC方法  

GFC method for Face Recognition

作　　者：刘宇[1] 肖沙里[1] 袁浩[1] 徐智敏[1] 汪科[1] 

机构地区：[1]重庆大学光电技术及系统教育部重点实验室

出　　处：《微型电脑应用》2006年第6期15-16,22,共3页Microcomputer Applications

摘　　要：Gabor-Fisher分类器(GFC)人脸识别方法,首先对面部图像进行Gabor小波处理,再对得到的增广Gabor特征向量应用增强线性辨别法模型(EFM)以得到辨别特征。该方法的新颖之处在于(1)引入了Gabor小波得到增广的特征向量更好的反映了图像的特征;(2)应用EFM在对维数进行降低的同时进行分类特征提取。该种新方法在应用于光照和面部表情变化比较大的FERET数据库时优势比较明显。The Gabor-Fisher Classifier (GFC) method for face recognition applies the Enhanced Fisher linear discriminant Model (EFM) to an augmented Gabor feature vector derived from the Gabor wavelet representation of face images. The novelty of this method comes from the derivation of an augmented Gabor feature vector, whose dimensionality is further reduced with the use of the EFM by considering both data compression and recognition performance. The new GFC method has more advantages for face recognition acquired under variable illumination and facial expressions.

关 键 词：GABOR小波 增强线性辨别法模型 Gabor-Fisher分类器 人脸识别 

分 类 号：TP139.4[自动化与计算机技术—控制理论与控制工程]