Abstract: In order to improve the practicability and recognition rate of face recognition system, this paper proposes a new method of feature extraction based on fusion of the Gabor Wavelet and Gist feature. This paper extracts the Gabor feature of face images from multiple scales and directions, obtains the Gist feature of those Gabor feature maps and then uses them as the final feature vector for subsequent recognition. Finally, this paper uses SVM as classifier and tests this new method in ORL and FERET face database respectively. Compared with the traditional method PCA-SVM, the results show this method significantly improves the face recognition rate.

Key words: face recognition, Gist feature, Gabor feature, feature extraction, Support Vector Machine（SVM）

摘要： 为了进一步增强人脸识别系统的实用性，提高人脸识别率，提出了一种新的融合Gabor小波特征和Gist特征的人脸特征提取方法。对一幅人脸图像提取其多个尺度和方向的Gabor特征图，再对这些Gabor特征图进行处理，分别提取其Gist特征，接着再把所有Gabor特征图的Gist特征级联起来作为一人脸图像的特征向量，经过PCA方法降维处理，最后输入到支持向量机里面训练识别。通过在ORL和FERET人脸库中进行实验检测，结果表明与传统的PCA-SVM方法和Gabor特征提取方法相比，给出的方法可以大幅度提高人脸识别率。

关键词: 人脸识别, Gist特征, Gabor特征, 特征提取, 支持向量机