Planar multi-classifier modelling-NN/NN for face recognition

In this paper we present an approach of face recognition by using a planar multi-classifier modelling based on neural networks. Each class of the database has been modelled by a PMCM-NN/NN formed by three secondary models and a main one - of an MLP type. We have retained the same characteristics and database used in [2]. The various experiments have been proven on a FERET face database. The recorded results have shown the contribution of our proposed concepts for the generalization of planar models, initially based on HMMs, to other types of classifiers - neural in this case. The validation of the PMCM in the case of face recognition shows results that are very promising and better than those registered in [2].