Biometric recognition suffers from time-span induced accuracy degradation
In many biometric systems the time of enrollment can be substantially separated from an actual authentication attempt (e.g., data as stored in e-passports remains valid for 10 years). Therefore, ageing might affect all parties involved in such an authentication process. While for some biometric modalities like face and speech such effects have been clearly demonstrated and techniques are established how to cope with these, the debate about these issues is quite controversial for iris and fingerprint recognition systems. We will discuss some recent results for iris and fingerprint template ageing and will show how to isolate certain reasons for observed time-span induced accuracy degradations.
Andreas Uhl is a professor for Computer Science at the University of Salzburg where he leads the Multimedia Signal Processing and Security Lab. His main research interests are in vision-based topics often closely related to security, e.g. multimedia security (e.g. visual data encryption, watermarking) and forensics, as well as biometrics. In the latter field he focuses on questions of biometric system robustness and security mainly considering the fingerprint, iris, and vein modalities. Besides conducting national projects, Andreas Uhl is currently involved in two biometrics-oriented Horizon 2020 projects, i.e. Pervasive and UseR Focused BiomeTrics BordEr ProjeCT (PROTECT) and Computer Vision Enabled Multimedia Forensics and People Identification (IDENTITY). In the academic field, Andreas Uhl was area chair for ocular biometrics @ International Conference on Biometrics (ICB'16) and is Program co-Chair of the annual International Conference of the Biometrics Special Interest Group (BIOSIG) since 2015.