Fingerprint biometrics from newborn to adult : a study from a national identity database system

Abstract

In this work, we evaluate the use of fingerprints to identify people from a very young age. Although it is well known that fingerprints are stable all along life, and even before born fingerprint patterns are fully developed, automatic identification (or comparison) systems are developed generally for adult fingerprints. Our interest is not only to study the feasibility of using child fingerprints for automatic identification but to determine if that is possible with the existing software and hardware. There are two related questions that we do answer in this work. First, starting at what age are digitally acquired fingerprints good enough for automatic comparison. Second, what is the performance when comparing fingerprints from children against fingerprints from adults. In order to answer these questions, we have run a set of experiments on a database composed of more than 200K fingerprints from approximately 134K identities. We show that, after applying a growth factor to scale minors fingerprints to an adult size, good accuracy can be obtained from ages starting at one year old, and that fingerprints of children and adults can be compared without a significant loss of accuracy (with respect to adult vs adult). We consider this study extremely useful for both researchers and decision makers, as it is a testimony that even without additional developments, fingerprints from children can be used for automatic comparison on real scenarios.