Unsupervised discriminant projection (UDP) has a good effect on face recognition problem, but it has not made full use of the training samples' class information that is useful for classification. Linear discrimination analysis (LDA) is a classical face recognition method. It is effective for classification, but it can not discover the samples' nonlinear structure. This paper develops a manifold-based supervised feature extraction method, which combines the manifold learning method UDP and the class-label information. It seeks to find a projection that maximizes the nonlocal scatter, while minimizes the local scatter and the within-class scatter. This method not only finds the intrinsic low-dimensional nonlinear representation of original high-dimensional data, but also is effective for classification. The experimental results on Yale face image database show that the proposed method outperforms the current UDP and LDA.
[1]
Zhongliang Jing,et al.
Local structure based supervised feature extraction
,
2006,
Pattern Recognit..
[2]
Xiaofei He,et al.
Locality Preserving Projections
,
2003,
NIPS.
[3]
Mikhail Belkin,et al.
Laplacian Eigenmaps for Dimensionality Reduction and Data Representation
,
2003,
Neural Computation.
[4]
Jian Yang,et al.
Locally principal component learning for face representation and recognition
,
2006,
Neurocomputing.
[5]
Xiaolong Teng,et al.
Face recognition using discriminant locality preserving projections
,
2006,
Image Vis. Comput..
[6]
Jian Yang,et al.
Globally Maximizing, Locally Minimizing: Unsupervised Discriminant Projection with Applications to Face and Palm Biometrics
,
2007,
IEEE Transactions on Pattern Analysis and Machine Intelligence.