Improving fingerprint matching using an orientation-based minutia descriptor

Abstract

This paper reviews a well-known fingerprint matching algorithm that uses an orientation-based minutia descriptor. It introduces a set of improvements to the algorithm that increase the accuracy and speed, using the same features. The most significant improvement is in the global minutiae matching step, reducing the number of local matching minutiae and using multiple minutiae pairs for fingerprint alignment. We conduct a series of experiments over the four databases of FVC2004, showing that the modified algorithm outperforms its predecessor and other algorithms proposed in the literature. © 2009 Springer-Verlag Berlin Heidelberg.