Since the terrorist attacks of 11 September 2001 in the USA interest in biometric identification systems has once again risen sharply among law enforcement agencies. In connection with the Federal Ministry of the Interior’s package of antiterrorism measures, the use of biometric identification to authenticate immigrants and visa applicants (Immigration Act, Prevention of Terrorism Act) is under consideration. Authentication in this context refers to the unmistakable link between an individual and his or her identity. This identity can be stored on a document available for production, whose authenticity can be verified, or it can be in a central repository. Where it is stored is immaterial. The link can be evidenced by distinguish-ing physiological or behavioural characteristics of the person – biometric features – that are captured and verified. The characteristics to be captured can be of a static nature, for example, fingerprints, hand geometry etc., or they can describe dynamicattributes such as the voice.
In the subprojects BioFace I and BioFace II, a comparative study of the recognition perform-ance of facial recognition systems was carried out. The studies were conducted firstly at the level of pure algorithm tests (laboratory tests) in the area of verification (1:1 comparison) and identification (1:n comparison) and secondly at the level of a test under realistic conditions of use in the area of identification (practical tests/system test). The primary aim was to analyse the capability of the systems with large volumes of data and the influence of noise factors. The present report documents the framework conditions, the data material used and the pro-cedure and results of the studies themselves.
Download the Report v21. (PDF) (6 MB) (Document may not be accessible to people with disabilities)