A Digital Geometry-Based Fingerprint Matching Technique

Fingerprint authentication systems have become ubiquitous with the event of smart devices and cheap sensors. Because of its use in a wide range of devices, it is important that the biometric authentication systems are computationally cheap, and at the same time robust to distortions and noises. Most of the existing fingerprint authentication systems are based on conventional mathematics or complex filter and consequently features extracted are computationally expensive or gross approximate with a large margin of errors leading to drops in the accuracy of the system. In this paper, we propose a fingerprint template based on translation and rotation invariant features of the delaunay triangulation of minutiae. However existing methods of comparison of delaunay triangles features do not always result in an optimum match because of the distortion in the different prints of the same finger. In the proposed method, features extracted from the delaunay triangulation are augmented with the information of local ridges based on discrete curvature and digital straightness. Experimentation with FVC2000 shows that the proposed method has better performances compared to the latest techniques.