Speaker: Mr. Nawaf alharbi

Title: Cancellable fingerprint Templates and Invariants for Fuzzy Authentication

Abstract:
We examine new methods for describing cancellable fingerprint templates, with associated matching algorithms. Cancellable fingerprint templates are examples of biometric authentication mechanisms, which have the additional property of protecting the identity of users between databases.
We use an existing algorithm to extract minutiae from fingerprint images. These minutiae represent the position and local neighborhood of discontinuities in the ridge flow of the fingerprints.
We have developed three methods of transforming the minutiae into cancellable template features, the line-based method, circle-based method, and the quad-method. These simple methods are designed to test the invariance dependence on the base parameters identified in minutiae: x-position, y-position and relative angular position. For each we have defined an appropriate matching algorithm.
We have carried out experiments to test the methods. We have applied each transform to a database of 80 images and measured the false positive and false negative rates. We have reported on the significance of the results as regards the use of the primitive minutiae parameters in constructing more complex invariants.