基于二维PCA的人脸识别方法研究

Research on Face Recognition Based on Two-dimensional PCA

该文提出了一种基于二维PCA的类内平均脸方法进行人脸的特征提取。首先利用类内平均脸对人脸训练样本进行规范化处理,根据规范化之后的人脸训练样本计算图像协方差矩阵,并求解一组最优特征向量,然后将人脸样本投影到这组最优特征向量上来提取人脸的特征,最后采用最近邻距离分类器来分类所提取的特征。此方法在NUST603人脸图像库上进行了实验,验证了该方法的有效性。

In this paper, a method based on within - class average faces of two - dimensional PCA is proposed for face featttre extraction. First, face training samples are normalized by using the corresponding within - class average face images. According to the normalized face training samples, the image covariance matrix is calculated to obtain a family of the optimal feature vectors. Then the face samples are projected onto the family of the optimal feature vectors to extract the face features. Finally, a nearest neighbor distance classifier is employed to classify the extracted features. Experimental results on the NUST603 face database show the method in this paper is effective.