基于PCA的人脸识别系统的研究与实现

摘要

人脸识别技术是人脸识别系统以人脸图像作为识别身份的媒介.研究了PCA算法的原理,基于matlab平台实现了人脸识别系统.该系统在对图像进行预处理的基础上,以ORL人脸库中部分图像为样本,采用K-L变换计算训练样本特征值及特征向量,得到特征脸向量,进而计算测试样本投影到特征脸空间的坐标系数,最后分类识别得出结果.实验证明该系统识别率较高,对实际应用有一定的参考价值.%The PCA algorithm theory was studied and the face recognition system was realized based on Matlab. Part of ORL face database was taken as samples. On the basis of the image pretreatment, feature values and the training sample feature vectors were computed with KL transform. Then, coordinate coefficients were calculated after the test samples were projected to the eigenface space. The final classification and recognition outcome was obtained. Experiments show that the recognition rate of the system is higher, and the method has some reference value for the practical applications.