利用Gabor小波变换解决人脸识别中的小样本问题

Solution of small sample size problem in face recognition using Gabor wavelet transform

提出了一种在人脸识别中解决小样本问题的新算法。通过把人脸图像经过Gabor小波变换后得到的每个图像都看成是独立的样本,大大增加了每一类人脸样本的样本数,解决了人脸识别中的小样本问题。专门针对人脸特征向量组,设计了使用白化变换后余弦距离测度的最近邻分类器来进行分类判决。在FERET人脸库中,对该方法与直接PCA方法进行了实验比较,结果表明,新方法的平均正确识别率可以达到97%,比直接PCA方法具有更好的识别性能。

A novel algorithm for solving the small sample size problem in face recognition is proposed. The size of each face sample is increased greatly and the small sample size problem is resolved by regarding every Gabor wavelet transformed output image as an independent sample. A special nearest neighbor classifier using whitened cosine distance is designed for the classification of face feature vector groups. The experiments comparing the prosposed approach with the direct PCA method in FERET face database show that the average correct face recognition rate of the proposed algorithm can reach 97%, which is superior to the direct PCA method.