摘要
在光照变化条件下,人脸识别的正确率急剧下降,为了解决该难题,提出了一种离散余弦变换和主成分分析相融合的光照变化条件人脸识别方法。首先对人脸图像进行分块,并采用离散余弦变换对每一个子块提取DCT系数,然后采用主成分分析提取人脸特征,并采用深度学习算法建立人脸识别的分类器,最后采用ORL和Yale B人脸库进行仿真实验,测试其有效性和优越性。实验结果表明,相比其它光照人脸识别方法,本文方法提高了光照人脸图像的识别率,消除了光照变化的不利影响,具有较强的鲁棒性。
In the illumination Change conditions,face recognition correct rate decreased sharply,in order to solve this problem,this paper proposed a face recognition method in illumination condition based on discrete cosine transform and principal components analysis.Firstly,the face images are divided into blocks,the DCT coefficient is extracted by discrete cosine transform for each sub block,and then uses the principal components analysis is used to extract face features,and the depth learning algorithm is used to f human recognize face,finally,the simulation experiments are carried out to test the validity and superiority by using multi person face database.The experimental results show that,compared with other illumination face recognition methods,this method has a higher rate of face recognition,eliminates the adverse effects of illumination changes,and has strong robustness.
出处
《激光杂志》
北大核心
2015年第4期126-130,共5页
Laser Journal
基金
陕西省教育厅科研计划项目(2013JK1195)
关键词
光照条件
离散余弦变换
特征提取
浓度学习
主成分分析
illumination condition
discrete cosine transform
features extraction
principal component analysis
deep learning