基于分块PCA的人脸识别

Face Recognition Based on Block-PCA

设计了一种基于分块主分量分析(Block-PCA)的人脸识别方案。预处理阶段,首先将一幅脸像按不同方位划分为大小相同的数个子块,对各子块进行能量归一化和傅氏变换,以消除部分光照影响并估算子块的频谱。在此基础上进行分块PCA,提取各子块主分量特征,再将子块的主分量特征整合为整体特征,最后采用最近邻判决准则进行分类识别。对ORL人脸数据库的实验结果表明所设计方案是有效的。

This paper presents a scheme for face recognition,based on block principal component analysis（Block-PCA） wherein preprocessing is conducted by partitioning each face image into several blocks with the same size,followed by energy normalizing to reduce the brightness variation effect and by the Fourier transform to estimate the spectra.Eigen features are acquired by performing PCA on the spectra of each block,and then combined into global features,from which the recognition results are obtained by the Nearest Neighbor（NN） classifier.Experimental results of the Olivetti Research Laboratory（ORL） face database show that the proposed method is feasible for face recognition.