Quaternion K-L transform and Biomimetic pattern recognition approaches for color-face recognition

This paper proposed a method of quaternion K-L transform and biomimetic pattern recognition (BPR) for color face recognition. The BPR aimed at optimal covering in the feature space Rⁿ using some complex geometric bodies for cover samples distribution in Rⁿ approximately in order to ¿recognize¿. We used quaternion K-L transform to extract the Eigen-faces of training samples and algebraic feature components of each sample for training by BPR. The method was applied to the face database ¿faces94¿ in Essex University, and the experiments results indicated that the correct recognition rate reached to 96.67%, and proved its efficiency and feasibility for color face recognition, and also showed it was better than the performance of SVM.