Artificial neural network based classification technique for minutiae verification

Abstract

The theory behind the fingerprint verification based on minutiae matching, was in detail studied. The performance of the developed system was evaluated on database with 2 fingerprints from 20 different people. The test showed that the system and algorithm is fully capable of distinguishing the related fingerprints apart from the non-related fingerprints. The system has proved to be robust towards translation, rotation and/or missing minutiae between the matched fingerprints. Future work can be improved by applying the fingerprint classification and recognition for more damaged and noisy fingerprint image database. System performance could be determined for database of more than 1000 images.

Keywords: Minutiae, artificial neural network, classification, fingerprint.