An Experimental assessment of the performance of Linear and Kernel-based Methods for Face Recognition

Abstract

This paper presents the results of a comparative study of linear and kernel-based methods for face recognition. These experimental results include: (1) a comparative study of linear methods for feature extraction, such as Principal Component Analysis (PCA), Fisher’s Linear Discriminant Analysis (FDA), and kernel based methods for feature extraction, such as Kernel based Principal Component Analysis (KPCA), Kernel based Discriminant Analysis (KDA). (2) a comparative study of linear methods for recognition or classification, such as Nearest Neighbor (NN), Linear Support Vector Machine (LSVM), and kernel based methods for classification, such as Kernel based Nearest Neighbor (KNN), Nonlinear Support Vector Machine (NSVM). In addition, we also obtain some interesting conclusions after all of these methods are performed on several well-known face database, i.e. ORL, YALE and UMIST Face Database, respectively.

6 Figures and Tables

Cite this paper

@inproceedings{Lu2004AnEA, title={An Experimental assessment of the performance of Linear and Kernel-based Methods for Face Recognition}, author={Congde Lu and Taiyi Zhang and Wei Zhang and Yulei Chen}, year={2004} }