Introduction

Abstract

Face recognition (Samal and Iyengar, 1992; Chellappa et al., 1995; Daugman, 1997; Jain et al., 1999; Zhao et al., 2003; Belle et al., 2004; Li and Jain, 2005) has become a major biometric technology. Biometrics involve the automated identification or authentication from personal physical appearance or behavioral traits. Human physical appearance and/or behavioral characteristics are counted as biometrics as long as they satisfy requirements that include universality, distinctiveness or uniqueness, permanence or invariance, collectability, and acceptability (Clarke, 1994). The early belief in the uniqueness aspect of faces (to preempt forgeries) was one of the reasons behind their use, e.g., the face of Queen Victoria on the early stamps (Samal and Iyengar, I992). Biometric systems, including face recognition systems, can be categorized according to their intended applications. According to Wayman (I999) a suitable self-evident taxonomy will include cooperative vs. non-cooperative, overt vs. covert, habituated vs. non-habituated, attended vs. non-attended, standard vs. non-standard operating environments, and public vs. private.