Face Gender Recognition Using Neural Networks and DCT Image's Coefficient Selection
Marco Grassi, Marcos Faúndez-Zanuy, Ondřej Šmirg, Honza Mikulka · Frontiers in artificial intelligence and applications · 2011
Gender recognition plays an important role for a wide range of application in the field of Human Computer Interaction. In this paper, we propose a gender recognition system based on 2D Discrete Cosine Transform and Neural Networks. In particular, a discriminability criterion is used to select the DCT coefficients that make up the biometric template. Experimental results show how the proposed approach leads to a significant enhancement of recognition performances.