2DPCA vs. 2DLDA: Face Recognition Using Two-Dimensional Method

Two-dimensional principal component analysis (2DPCA) and two-dimensional linear discriminant analysis (2DLDA) are new techniques for face recognition. The main ideas behind 2DPCA and 2DLDA are that they are based on 2D matrices as opposed to the traditional PCA and LDA, which are based on 1D vector. In some literature, there has been a tendency to prefer 2DLDA over 2DPCA because, as intuition would suggest, the former deals directly with discrimination between classes, whereas the latter deals with the data in its entirely for the principal components analysis without paying any particular attention to the underlying class structure. In this paper, to compare the performances of the two methods, a series of experiments perform on two face image databases: ORL and CAS-PEAL. The experiments results show that the performance of 2DLDA is not always better than that of 2DPCA. Particularly, in the case of large subjects, 2DPCA can outperform 2DLDA.