We propose a combined feature Fisher classifier method for face recognition (FR). Both facial holistic and local information are used for face representation. In addition, the enhanced Fisher linear discriminant model is employed for improving the generalization capability. Experiments show that the method is not only robust enough to moderate changes of illumination, pose, and facial expression but also superior to the traditional FR methods, such as eigenfaces and Fisherfaces.
"Enhanced Fisher linear discriminant model based face recognition using facial combined feature," Optical Engineering 43(12), (1 December 2004). https://doi.org/10.1117/1.1812539