Bi-Objective Optimal Image Discriminant Analysis and Its Application in Face Recognition

The paper concerns the image discriminant analysis problem where image matrices are taken as samples.Foley-Sammon discriminant analysis is transformed into a class of bi-objective constrained optimization problem,and the efficient projection vector is defined.By the necessary conditions for multi-objective optimization,we shown that the efficient projection vector is the eigenvector of eigen-equation corresponding to the largest eigenvalue.As a result,a method to find the set of efficient projection vectors is obtained. Here,the non-singularity of the within-scatter matrix is not essential.The experiments show that the computational time could be greatly reduced if our proppsed method is used for feature extraction,and the performance of its recognition is superior to the others.