摘要
为提高人脸识别的准确率,缩短图像特征提取的时间,提出了一种将二维主成分分析(简称2DPCA)与改进的线性鉴别分析(简称LDA)相结合的人脸识别方法。该法首先以图像矩阵为分析对象,直接利用原始图像矩阵构造图像的协方差矩阵,以进行特征提取和2DPCA分析;再采用改进的线性鉴别分析,得到最佳的分类特征,从理论上有效解决了传统的线性鉴别分析在人脸识别中存在的"边缘类"问题;最后,在ORL人脸库上检验了该识别方法的性能。实验结果表明,该方法抽取的鉴别特征有较强的鉴别能力。
A human face recognition technique based on Two-Dimension Principal Component Analysis （2DPCA） and Linear Discrimination Analysis is presented. With this method, first, the original image matrix is directly utilized to construct covariance matrix for feature extraction. Compared with Principal Component Analysis （PCA）, it not only improves correct recognition rate in face recognition, but also reduces much more time in feature extraction; and then a improved Linear Discrimination Analysis is applied to obtain classification feature. This improved method gives an effective way to resolve the ＂edge of class＂ problem of the traditional Linear Discrimination Analysis theoretically in face recognition. Finally, extensive experiments performed on ORL face database verifies the effectiveness of the proposed method based on 2DPCA and improved Linear Discrimination Analysis.
出处
《南通职业大学学报》
2009年第3期88-92,共5页
Journal of Nantong Vocational University
