基于对称最大间距准则的人脸识别方法

摘要

A symmetry based maximum margin criterion ( SMMC) on the basis of this idea of symmetry is presented , which was introduced into maximum margin criterion ( MMC) . Firstly, this algorithm is based on the theory of function decomposition and mirror transform. The original face samples are divided into the even symmetrical images and odd symmetrical ones. Then MMC was performed in the even symmetrical images and odd symmetrical images for feature extraction respectively. Experimental results on ORL face database and Yale face database show that the algorithm enlarged sample capacity by using the mirror transform. Therefore, SMMC not only expresses the advantages of MMC, but also remarkably improves die recognition rates.%在最大间距准则算法中引入对称性思想,提出了基于对称最大间距准则(SMMC)的人脸识别算法.该算法首先引入镜像变换,将人脸图像生成镜像图像.依据奇偶分解原理,把人脸图像分解成镜像奇、偶对称图像.然后分别对奇偶对称图像进行最大间距准则,提取所需的鉴别特征.在ORL和Yale标准人脸库上的实验结果表明,该算法通过利用镜像变换扩大了样本容量,发挥了最大间距准则算法的优点,提高了人脸识别率.