Biometric authentication based on multi-instance fingerprint fusion in degraded context
Rzouga Haddada Lamia, Essoukri Ben Amara Najoua · 2019
Fingerprint authentication has been a trending research topic for a long time, with many applications. It is one of the most widespread methods for criminal investigations. Yet, in such cases, fingerprint images from multiple fingers are degraded, which keeps their use very challenging. In order to design a reliable fingerprint authentication system, we propose multi-instance fingerprint fusion in a degraded condition. We validate the suggested approach on the most frequently deployed three fingers of the right hand: the thumb, the index and the major. The fusion of multiple fingerprints is made by the Choquet integral at the score level using particle swarm optimization to set their fuzzy measures. The experimental results show that the proposal offers a compromise between accuracy and cost and yields a low equal error rate of 0.22%.