In this paper we describe a 3d face recognition system based on neural networks. The system consists of a modular architecture in which a set of probabilistic neural networks cooperate with the associated graphical models in recognising target people. The logic of this cooperation is quite simple: each network is able to discriminate between its “target” and all other samples of the training set. This is done by using only one characteristic piece of information among the available sets of L,U, V colors and Z coordinate. Every network provides its associated graph with estimates obtained during the training phase, while graphical models coordinate the answers of all the associated networks giving the posterior probability that the target corresponds to the person to be recognised. Then a decision-making criterium based on the maximum posterior probability is established to identify the recognised face
Mauri, G., Zoppis, I. (2003). A probabilistic neural networks system to recognize 3D face of people. In WIRN 2003 - 14th Italian Workshop on Neural Nets, Revised Papers (pp.158-164). Springer Verlag [10.1007/978-3-540-45216-4_17].
A probabilistic neural networks system to recognize 3D face of people
MAURI, GIANCARLO;ZOPPIS, ITALO FRANCESCO
2003
Abstract
In this paper we describe a 3d face recognition system based on neural networks. The system consists of a modular architecture in which a set of probabilistic neural networks cooperate with the associated graphical models in recognising target people. The logic of this cooperation is quite simple: each network is able to discriminate between its “target” and all other samples of the training set. This is done by using only one characteristic piece of information among the available sets of L,U, V colors and Z coordinate. Every network provides its associated graph with estimates obtained during the training phase, while graphical models coordinate the answers of all the associated networks giving the posterior probability that the target corresponds to the person to be recognised. Then a decision-making criterium based on the maximum posterior probability is established to identify the recognised face
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
