A Hybrid Approach to Face Recognition Using Generalized Two-Dimensional Fisher's Linear Discriminant Method

Abstract

This paper presents a novel Hybrid Approach to Face Recognition using Generalized Two-Dimensional Fisher’s Linear Discriminant (HAFR G-2DFLD) Method. It has been seen that the facial changes due to variations of pose, illumination, expression, etc. are appeared only some regions of the whole image. Therefore, the conventional face recognition methods, which use whole image for feature extraction and recognition, do not result much success. To cope with the above facial changes, face images are divided into a number of non-overlapping sub-images and then G-2DFLD method is applied to each of these sub-images as well as to the whole image to extract local and global discriminant features, respectively. A multi-class SVM is used as a classifier for each of the sub-images and also for the whole image for recognition based on those extracted features. Finally, a decision is made for recognition of the image by fusing the decisions of the individual SVMs. The proposed HAFR G-2DFLD method was evaluated on two popular face recognition databases, the AT&T and the UMIST face databases. The experimental results show that the new HAFR G-2DFLD method outperforms the conventional global feature extraction methods like, PCA, 2DPCA, PCA+FLD, 2DFLD and G-2DFLD methods in terms of face recognition.