Online biometric authentication using facial thermograms

Abstract

The requirement for a reliable personal identification in computerized access control, security applications, human machine interaction etc has led to an unprecedented interest in biometrics. The usefulness of face as a primary modality for biometric authentication is on the rise in the recent years because of it's non-intrusiveness and uniqueness. Visual Face recognition is successful only in the controlled environment but fails in the case of disguised faces and under varying lighting conditions. As an alternative to the visual recognition this paper presents the Long Wave Infra Red (LWIR) for face recognition. In this we make of the facial thermograms that are the images formed by the capturing the heat radiated by the face. It is observed that it's performance falls drastically with varying temperature conditions. To overcome this drawback simplified Blood perfusion model is proposed to convert thermograms into Blood perfusion data. If a person wears spectacles, the glasses obstruct the radiated and hence thermograms loses the information. An efficient algorithm is developed to detect the eyeglasses and to remove it's effect.