Optimal weight selection in matching score fusion based Face Recognition

Face recognition is an active subject in the fields of pattern recognition. There still exist research difficulties in its practical application in the term of performance accuracy. However, the above limitation can be overcome by combining the biometric information extracted from diverse facial features such as total face, eyes, nose, mouth, etc. In this paper we present a Face Recognition system in which a new Bees Algorithm based optimal weight searching method is used to find the optimal parameters to fuse the aforementioned information at matching score level. The proposed method tested on the ORL face database achieves 99% recognition rate, which is overall superior to those traditional FR approaches such as the Eigenfaces, Fisherfaces, D-LDA, IFS methods.