基于Gabor的伸长局部二值模式及其人脸识别应用

A method based on Elongated local Gabor binary pattern applied to face recognition

提出一种基于Gabor的伸长局部二值模式(elongated local binary pattern,ELBP)的人脸识别方法。该方法首先对人脸图片进行Gabor滤波,得到一组Gabor幅值图像(Gabor magnitude maps,MMPs);然后利用ELBP提取每一幅幅值图像的纹理特征,并用ELBP纹理特征的直方图特征和平均最大距离梯度幅值特征联合表示该图像纹理特征;最后,通过比较测试图片和训练集的直方图交叉距离进行识别。在YALE,YALE-B,UCD-VALID,CMU-PIE等人脸库进行测试,所提方法取得了理想的效果,证明所提方法能够有效地进行人脸识别。

In order to improve the face recognition rate,a method based on local Gabor binary pattern for face recognition is presented.Convolving the face image with Gabor filters to gain a group of magnitude maps（MMPs）,and then extracting histogram of each magnitude map＇s ELBP to indicate its textural features,uniting with Average Maximum Distance Gradient Magnitude.At the end,comparing the histograms of test images and training set through histogram crossing distance to identify people.Experiments are carried out on YALE、YALE-B、UCD-VALID、CMU-PIE face databases,and experimental results have validated the effectiveness of the proposed method.The results prove that the proposed method can effectively improve the face recognition rate.