基于PCA和SVM的人脸识别系统  

Face recognition system based on PCA and SVM

作　　者：胡沐晗 Hu Muhan(Department of Information, ZhongBei College of NanJing Normal University, Nanjing, Jiangsu 210046, China)

机构地区：[1]南京师范大学中北学院信息系,江苏南京210046

出　　处：《计算机时代》2017年第12期60-63,67,共5页Computer Era

摘　　要：PCA(主成分分析)与SVM(支持向量机)技术对于人脸识别而言非常重要,文章结合PCA与SVM来设计一个人脸识别系统。在MATLAB编译环境下,分别探讨PCA技术与SVM技术,总结出一种快速PCA算法以及SVM的测试与训练算法;从orl_faces人脸库中选取训练数据与测试数据进行训练与测试,对其中重要的参数进行调优,最终达到90%的识别成功率。再经真实人脸的训练与测试,达到了73%的真实人脸识别成功率。PCA(Primary Component Analysis)and SVM(Support Vector Machine)technologies are very important for face recognition.This paper combines PCA and SVM to design a face recognition system.In the MATLAB compiler environment,the PCA technology and SVM technology are discussed,a fast PCA algorithm is summed up as well as the test and training algorithm for SVM;the training data and test data are selected from the orl_faces database to train and test,the important parameters are optimized,and90%recognition success rate is ultimately achieved.After the training and testing by real face,the success rate of73%real face recognition is achieved.

关 键 词：PCA技术 SVM技术 人脸识别 训练数据 测试数据 MATLAB 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]