School of Science
Biometric security has found many applications in Internet of Things (IoT) security. Many mobile devices including smart phones have supplied fingerprint authentication function. However, the authentication performance in such restricted environment has been downgraded significantly. A number of methods based on Delaunay triangulation have been proposed for minutiae-based fingerprint matching, due to some favorable properties of the Delaunay triangulation under image distortion. However, all existing methods are based on 2D pattern, of which each unit, a Delaunay triangle, can only provide limited discrimination ability and could cause low matching performance. In this paper, we propose a 3D Delaunay triangulation based fingerprint authentication system as an improvement to improve the authentication performance without adding extra sensor data. Each unit in a 3D Delaunay triangulation is a Delaunay tetrahedron, which can provide higher discrimination than a Delaunay triangle. From the experimental results it is observed that the 3D Delaunay triangulation based fingerprint authentication system outperforms the 2D based system in terms of matching performance by using same feature representation, e.g., edge. Furthermore, some issues in applying 3D Delaunay triangulation in fingerprint authentication, have been discussed and solved. To the best of our knowledge, this is the first work in literature that deploys 3D Delaunay triangulation in fingerprint authentication research.
Available for download on Wednesday, May 01, 2019