The Impact of Template Aging on the Performance of Automatic Fingerprint Recognition

Abstract

It is generally accepted that fingerprints do not change with age. The longstanding practice of using fingerprints for person recognition confirms this thesis, since it has been used successfully for over a century. However, in the past, person recognition based on fingerprints has been conducted only by forensic experts, but today, especially with the introduction of biometric documents, automatic person-recognition systems are in common use. But do fingerprints have enough long-term stability for reliable automatic person recognition, even with a time interval between the template- and the input fingerprint-recordings over a span of more than a decade? This paper explores the phenomenon of fingerprint template aging, i.e. an increase in the error rate of minutiae based automatic fingerprint identification systems (AFISs) recognition with increased time since fingerprint enrolment. Based on the results of statistical analysis of AFISs similarity matching scores of fingerprints pairs of all fingers of the right and left hand recorded over a span of 21 years, despite the fact that on average approx. 9% of the variability of the AFISs matching scores can be explained by template aging, we find that fingerprint template aging has a statistically significant negative impact on the performance of the AFISs, even on a relative young white-male population aged from 14 to 53 years.