Information Technology Journal1812-56381812-5646Asian Network for Scientific Information10.3923/itj.2013.1184.1191ZengJing KangZhiwei 62013126Face recognition is an important issue in the field of pattern
recognition, artificial intelligence and machine vision. The face image is not
only sensitive to environment, illumination, pose and expression variation,
but also has high-dimensional data disaster. In recently years, Sparse representation
has been extensively studied in pattern recognition, which surprisingly pointed
out that one target sample can be accurately represented as linear combination
of very few measurement samples. Sparse Preserving Projection is a newly developed
feature extraction method for reducing dimensions, which seeks a subspace where
such sparse reconstructive relations among all training samples are preserved.
However, Sparse Preserving Projection is time-consuming. In this study, a novel
Kernel-based Collaborative Preserving Projection method is proposed for feature
extraction. The method first uses a kernel-induced distance measure to determine
k nearest neighbors of the target training sample from the remaining training
samples and then the target training sample is reconstructed from its k nearest
neighbors using collaborative representation. Finally the method seeks a low-dimensional
subspace where the local collaborative reconstructive relations among all the
training samples can be preserved. Experimental results on three benchmark databases
show that the method can achieve a better classification result than Sparse
Preserving Projection and reduce computation complexity.]]>Belhumeur, P.N., J.P. Hespanha and D.J. Kriegman,1997Jing, X.Y., D. Zhang and Y.Y. Tang,2004He, X.F., S.C. Yan, Y.X. Hu, P. Niyogi and H.J. Zhang,2005Xu, Y., D. Zhang, J.Yang, Z. Jin and J.Y. Yang,2011Roweis, S.T. and L.K. Saul,2000Wright, J., A.Y. Yang, A. Ganesh, S.S. Sastry and Y. Ma,2009Qiao, L., S. Chen and X. Tan,2010Zhang, L., M. Yang and X. Feng,2011Li, C.G., J. Guo and H.G. Zhang,2010Mallat, S.G. and Z. Zhang,1993Donoho, D.L.,20061 -norm solution is also the sparsest solution.]]>Mika, S., G. Ratsch, J. Weston, B. Scholkopf and K.R. Mullers,1999Chen, S. and D. Zhang,2004