Neural network and optical correlators for infrared imaging based face recognition

Abstract

This paper deals with the problem of thermovision application for biometric people recognition. The main goal of research was the automatization of recognition process and proving, whether the infrared imaging can be exploited for realization of the biometric system. First, the data base composed of 270 images made in different conditions, was built. Next, the multilayer perception and two optical correlators such as joint transform correlator and 4f correlator, were applied. Each system was designed for realizing two various tasks: verification and identification. Finally, the systems were tested and evaluated according to the Face Recognition Vendor Tests (FRVT). Achieved results proved that temperature distribution on the human face is a biometric trait and people can be efficiently recognized with usage of infrared imagery.