Abstract: In order to extract the features of human faces effectively, a face recognition algorithm based on traditional PCA method is proposed in this article, which combined with gamma transform and wavelet transform. The face images are processed with gamma transform, which can eliminate the effects of light and other nonlinear factors. It decomposes the face images by wavelet transform. It uses the low frequency component to instead the original face. It extracts face feature from the low frequency component of face by PCA to get the final identification characteristics. When tested in ORL faces database, the recognition rate of this algorithm is 6.5% higher than traditional algorithm based on PCA.

Key words: face recognition, gamma transform, wavelet transform, principal component analysis

摘要： 为了有效地提取人脸特征，提出了一种在传统PCA算法的基础上，结合伽马变换与小波变换的人脸识别算法。该方法对人脸图像进行伽马变换，消除光照等非线性因素的影响；对变换后的人脸图像进行小波分解，用得到的低频分量来替代原始人脸；对得到的人脸低频分量作PCA特征提取，得到最终的鉴别特征。在ORL人脸库上进行测试，该算法的识别率比传统的PCA算法提高了6.5%。

关键词: 人脸识别, 伽马变换, 小波变换, 主成分分析