摘要
本文介绍了Gabor-fisher classifier分类器(GFC)人脸识别方法,利用该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,该方法对于人脸图像在光照和表情变化的情况下仍然是有效的。该方法的新颖之处在于:(1)引入了Gabor小波,得到的增广特征向量更好地反映了图像的特征;(2)应用EFM在对维数进行降低的同时进行分类特征提取。通过与传统的LDA方法和PCA方法的对比得出,该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 linear discriminant model (EFM) to an augmented for face recognition applies the enhanced fisher Gabor feature veetor derived from the Gabor wavelet representation of face images. The novelty of this method comes from (1) the derivation of an augmented Gabor feature vector, (2) Dimensionality is further reduced using the EFM by considering both data compression and recognition performance. The new GFC method has more advantags tested on face recognition which were acquired under variable illumination and facial expressions.
出处
《国外电子测量技术》
2006年第11期63-65,共3页
Foreign Electronic Measurement Technology
基金
系光电技术及系统重点实验室教育部访问学者资金支持项目(CQU0308)。