摘要： 为了克服光照变化较大的情况对识别率的影响，提出基于二阶双向二维主成分分析(Sec-(2D)2PCA)的人脸识别方法。丢弃提取人脸图像的(2D)2PCA的前几个反映光照信息的主成分。在剩余图像中再次使用(2D)2PCA方法。Yale人脸库B和Yale人脸库上的试验结果表明，该方法在识别性能上优于2DPCA、(2D)2PCA、Sec-2DPCA方法。

关键词: 特征提取, 主成分分析, 人脸识别

Abstract: In order to overcome the effect which great illuminative changes have on the recognition rate, a face recognition technique based on Sec-(2D) 2PCA is presented in this paper. It discards principal components of the first several reflective illumination information that are extracted from the original face images with (2D)2PCA method. And then, (2D)2PCA is used again in the surplus images. A series of experiments are performed on face image databases: Yale human face database B and Yale human face database. The experimental result indicates that recognition performance of Sec-(2D)2PCA is superior to that of 2DPCA, (2D)2PCA, Sec-2DPCA method.

Key words: feature extraction, principal component analysis, face recognition

中图分类号: 

• TP391.41