An Artificial Neural Network Based Approach to Mass Biometry Dilemma Taking advantage from IBM ZISC-036 Neuro-Processor Based Massively Parallel Implementation

Abstract

Over the recent past years, new public security tendency to fit up public areas with biometric devices has emerged new requirements in biometric recognition dealing with what we call here "mass biometry". Ifthe main goal in "individual biometry" is to authenticate and/or identify an undesired individual within a set of favored folk, the main goal in "mass biometry" is to authenticate and/or identify an unusual (suspect) behavior within a flow of mass customary behaviors. So, in "mass biometry ” the ability of handling patterns containing relatively poor information and the skill of high speed processing in order to treat a mass number of patterns in real-time are chief requirements. These antagonistic requirements make authentication and identification tasks very challenging for the "mass biometry " related applications. In this paper we present an Artificial Neural Network (ANN) based face recognition system in a "mass biometry" context using facial biometric features. The proposed system takes advantage from kernel functions based ANN model and its IBM ZISC-036 based massively parallel hardware implementation. Experimental results validating the issued prototype mass biometric system is been presented and discussed.