Fuzzy label canonical correlation analysis and its application to face recognition

Incorporating the sample distribution information into the process of feature extraction is beneficial to promoting the classification performance of features. A fuzzy label canonical correlation analysis (CCA) algorithm is proposed for image feature extraction. Fuzzy class labels in the form of membership degrees are designed elaborately to represent the sample distribution. Then the fuzzy labels are embedded in CCA to extract more discriminative features which combine the information about gray level and distribution together. Furthermore, according to the matrix theory and dual-space idea, an improved method named dual-space fuzzy label CCA is proposed to counteract the effect of small eigenvalues which are poorly estimated due to finite samples. The experimental results on ORL and combined face databases show that the features have a powerful ability of recognition, and that the proposed methods are efficient and practical.