Infrared facial recognition using modified blood perfusion

Abstract

An infrared (IR) facial image represents a measurement of thermal distribution of a face which is independent of illumination. However, it has been found that facial thermograms vary with ambient temperature, physiological as well as psychological conditions, which result in severe decline in recognition rate. In this paper, we present a modified blood perfusion model which is much simpler than the original one. The experimental results have illustrated that the performances of the two models are comparable and outperform the thermal data for same-session data. To be more important, it is demonstrated that performance of the modified blood perfusion model is superior to the original model for time-lapse testing. ©2007 IEEE.