摘要
为了使人脸识别算法能快速准确地识别人脸身份,提出了基于二维PCA与SVM算法相结合的人脸识别算法。在PCA算法提取特征的基础上,运用二维PCA算法对人脸进行特征提取。利用SVM多类分类器寻找人脸样本之间的最优分割超平面,对特征人脸进行训练和分类。对未训练的人脸进行二维PCA特征提取,再用SVM多类分类器进行识别,与训练的人脸库中的人脸进行匹配,确定人脸身份。实验结果表明,基于二维PCA和SVM算法的人脸识别系统具有更快的识别速度和更高的准确度。
In order to identify the face quickly and accurately,face recognition algorithm based on two dimensional PCA and SVM is proposed.Based on the characteristics of PCA algorithm,two-dimensional PCA algorithm is used to extract the feature of human face.SVM multiclass classifier is used to find the optimal split hyperplane between face samples,and training and classification of characteristic faces are carried out with SVM.The two-dimensional PCA feature extraction is carried out on the untrained face,then SVM multiclass classifier is used to identify and match the face in the trained face database to determine the face identity.The experimental results show that the face recognition system based on 2 DPCA and SVM has faster recognition speed and higher accuracy.
出处
《桂林电子科技大学学报》
2017年第5期391-395,共5页
Journal of Guilin University of Electronic Technology
基金
广西自动检测技术与仪器重点实验室主任基金(YQ14105)
广西科学研究与技术开发计划(桂科攻11107001-40)
关键词
人脸识别
二维PCA
特征提取
SVM
face recognition
two-dimensional PCA
feature extraction
SVM