针对人脸易受非均匀光照等因素而干扰其识别效果,提出了Gabor特征和深度信念网络(DAN)相结合的人脸识别算法.首先利用Gabor小波变换处理人脸图像得到Gabor特征;其次把得到的Gabor特征当作深度信念网络(DBN)的输入,并将网络进行逐层训练,进而在最高层构建其分类面;最后,对人脸图像样本进行识别基于训练好的深度信念网络.通过仿真,极大改善了人脸图像在非均匀光照干扰下的识别效果,识别率也有很大的提高,达到了很好的的鲁棒性.%This paper presents a face recognition algorithm which combines the features of Gabor and deep belief network (DBN).First,Gabor characteristics were obtained by using Gabor wavelet transform.Secondly,the obtained Gabor features were taken as the input to depth belief network (DBN),and the network was trained to build its classification surface at the highest level.Finally,the human face image sample was recognized based on the training good depth belief network.Through the simulation experiment,the recognition effect of the face images in the non-uniform illumination interference is greatly improved,and the recognition rate has been improved greatly,which has achieved good robustness.
展开▼
