Linear discriminant analysis for face recognition: comparison of subspace approach with regularization method
Daniel Grzywczak, Władysław Skarbek · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2013
This paper compares two face recognition methods based on Linear Discriminant Analysis: Subspace and Regularization method. These methods are tested for different parameters, different size of feature vector, Euclidean distance and modified Euclidean distance, for both LDA and Dual LDA method. Before using LDA, face images are described using selected frequency channels of DFT.