基于弧形Gabor小波的人脸识别

Face recognition based on the curve Gabor wavelet

针对Gabor小波提取人脸曲线脸部特征不够完整的问题,通过增加一个曲率参数,将常规的Gabor小波扩展为弧形Gabor小波(CGW),以获取更多有效的人脸特征。CGW在5个尺度和16个方向上提取人脸特征,利用主成分分析(PCA)对人脸特征进行选择和降维,并采用最近邻方法进行分类。在ORL数据库中每人选取5张图像作为训练样本,5张图像作为测试样本,并分别与PCA和Gabor小波方法进行比较。实验结果表明,CGW具有更好的识别效果。

In order to solve the problem of face feature incompletely extracted by Gabor wavelet,a curvature parameter is added to structure curve Gabor wavelet（CGW）for obtaining more effective face features.Face features are taken through CGW on 5scales and 16 directions.Then principal component analysis（PCA）method is used to select the features and reduce dimensionality.Finally the nearest neighbor method is used to classify.Five images for everyone in the ORL database is chosen to train,and the other five images is used to test.The experimental results show that CGW has better recognition.