In this paper, we combine the advantages of (2D)²PCA and (2D)²LDA, and propose a two-stage framework: "(2D)²PCA+(2D)²LDA". In the first stage, a two-directional 2D feature extraction technique, (2D)²PCA, is employed to condense the dimension of image matrix; in the second stage, the two-directional 2D linear discriminant analysis (2D)²LDA is performed in the (2D)²PCA subspace to find the optimal discriminant feature vectors. In addition, the proposed method can take full advantage of the descriptive information and discriminant information of the image. Experiments conducted on ORL and Yale face databases demonstrate the effectiveness and robustness of the proposed method.

Date Published: 8 July 2011
PDF: 4 pages
Proc. SPIE 8009, Third International Conference on Digital Image Processing (ICDIP 2011), 800934 (8 July 2011); doi: 10.1117/12.896278

Published in SPIE Proceedings Vol. 8009:
Third International Conference on Digital Image Processing (ICDIP 2011)
Ting Zhang, Editor(s)