Research on Face Recognition Method Based on Combination of SVM and LDA-PCA
In the face recognition, the nonlinear factors, such as the light, expression and gesture, has great changes, the recognition effect of the PCA algorithm has been seriously influenced. This paper proposes an LDA-PCA algorithm, that merged the idea of LDA (Linear Discriminant Analysis) into PCA algorithm, to obtain the eigen-face subspace and use of the LDA's idea to discriminate and analyze, and then select the feature face vector, mainly reflecting the category difference, to form a new subspace. Then taking use of SVM classifier on the new subspace. The simulation results, on the improved face database, show that the LDA-PCA algorithm can effectively improve the robustness of nonlinear factor and the face recognition rate.
KeywordsPrincipal component analysis (PCA) Linear discriminant analysis (LDA) Support vector machine (SVM) Face recognition
This research was supported by the Tianjin natural science Fund (13JCYBJC15800).
- 2.Peng H, Zhang C, Rong G et al (1997) Research of automated face recognition based on K-L transform. J Tsinghua Univ (Sci Tech) 3:67–70Google Scholar
- 3.Sun D, Li L, Wu L (2001) Face recognition feature space optimization method. Signal Process 17(6):510–514Google Scholar
- 5.Chen F, Chen X, Zhang S et al (2006) A human face recognition method based on modular 2DPCA. J Image Graph 11(4):580–585Google Scholar
- 6.Zhang L (2010) Face recognition method using improved modular 2DPCA. Comput Eng Appl 46(13):147–150Google Scholar
- 7.Sun Z, Wang F, Wu X (2012) Face recognition based on modular Kernel principal component analysis. Comput Digit Eng 40(8):119–121Google Scholar
- 8.Bian Z, Zhang X (2007) Pattern recognition. Tsinghua University, Beijing, pp 87–91, 222–226Google Scholar