An Iterative Algorithm for the Generalized Optimal Set of Discriminant Vectors and Its Application to Face Recognition

The generalized optimal set of discriminant vectors is a generalized version of Foley-Sammon optimal set of discriminant vectors. The main difference between the generalized optimal set of discriminant vectors and Foley-Sammon optimal set of discriminant vectors is that the separability of the projected set of the samples is considered from global view when calculating the generalized optimal set of discriminant vectors, that is, the projected set of the samples on the generalized optimal set of discriminant vectors have the best separability in global sense. This paper presents the definition of generalized optimal discriminant vectors, analyzes the shortcomings of the old algorithm, and puts forward a new iterative algorithm, which is proven theoretically to converge to the precise solution while the errors are also taken into consider. And lastly, the algorithm is applied to face recognition, the results of which show that it is more effective than the old method.