Automatic Recognition of Faces by Neural Networks and Principal Component Analysis
Hicham Tebbikh · 2005
In this paper, we propose a new hybrid method for the recognition of faces combining the neural networks with the principal component analysis. By using the geometrical approach, we carry out a preliminary classification of the faces by PCA before using a neuronal classifior (PMC). The results, compared with those of the PCA and PMC classifiors, give a clear improvement in term of classification. To carry out our tests, we used a very rich data base created in the Laboratory of Automatics and Informatics of Guelma LAIG . Mots cles : Geometrical approach, neural networks (NN), principal component analysis (PCA), classification, neural- PCA.