Biometrics for massive access control - traditional problems and innovative approaches

Abstract

The increase in security measures due to the complex international situation is forcing the realization of several access control systems equipped with biometric identifiers. Apart from technical problems, a number of nontechnical issues, strongly related to the user's acceptance, may have a strong influence on the design of the application and suggest particular choices in the selection of the biometric technique to be adopted. Starting from the assumption that facial recognition represents one of the most widely "accepted" biometric techniques, and that, due to contradictory performances shown in operating conditions, this methodology is seldom considered for high-security applications, the authors have investigated the possibility of improving the error figures by means of an original approach. The innovation consists in "enriching" the template obtained by means of canonical facial recognition algorithms with additional information extracted from behavioral characteristics of the user. The study, still in the early stage, is carried out in the framework of a collaboration between the National Research Council of Italy (CNR) and the Italian Ministry of the Defense (MoD) in the area of the strong authentication for physical access to military compounds. The aim of the present paper is only the highlighting of the potentiality of such an approach since first experimental data are not mature to generate consolidated error figures such as FAR and FRR. In any event, a very good capability in distinguishing between identical twins, a very well known problem in facial recognition, is a promising preamble for the continuation of the research.