Face recognition of technology experts for e-government based on web mining

Both Principal Component Analysis (PCA) and Two-dimensional Principal Component Analysis (2DPCA) are successful face recognition algorithm. High recognition accuracy can be achieved using these two methods on the normal facial database, such as ORL, FERET, Yale, AR, etc. In order to construct an enormous Face Database of Science and Technology Experts (FDSTE) for e-government, face recognition algorithm based on Web mining should be researched on pertinences. Because of the quality of images extracted from Web is largely distinct. So, an improved method based on 2DPCA called M-2DPCA is proposed in this paper which combines the merits of 2DPCA, PCA and LDA. That is, firstly, 2DPCA is used to deal with the original image matrixes, and then PCA is used to compress the feature matrixes again, at last we apply Fisher criterion to obtain the optimal set of discriminable vector for better classification. Experiments on the FDSTE show that M- 2DPCA is best than 2DPCA and PCA in terms of the classification speed and the recognition rate for the face recognition by Web mining. Experiment demonstrates that the proposed algorithm performance and proves this case is meaningful and useful for researching and application on the face recognition by Web mining.