基于PCA算法的人脸识别方法研究比较

A Comparative Study on Face Recognition Using PCA-based Methods

主成分分析(Principal Component Analysis,PCA)方法是人脸识别技术中一种广泛应用的数据降维技术。当通过使用PCA变换获得的主成分去重建原始人脸图像时,能使均方误差最小。在传统的PCA基础上,Yang等人提出了2DPCA方法,避免了从图像矩阵向一维向量的转换,并在人脸识别中获得了满意的效果。文章对这两种方法做了理论上比较并给予实验数据支持,实验证明,2DPCA在识别方面略优于传统PCA算法。

Principal Component Analysis is a popular method for dimension reduction widely used in face recognition. Based on the traditional PCA method, 2DPCA method is proposed which avoids the transformation from the 2D matrix to 1D vector,and has acquired good recognition accuracy. This paper compares these methods on theory and experimental data,and shows that the 2DPCA method is superior to the traditional PCA method.