Fingerprint Ridge Orientation Modeling

In the last decade, automatic fingerprint based personal authentication has matured to the point where it can be successfully applied in a myriad of applications. These applications are ranging from the low cost door opener for private use up to the critical government application ensuring national security. One common ground of all this fingerprint based personal authentication systems is the necessity for estimation of fingerprint ridge orientation. The importance of ridge orientation can be deflected from the fact that it is inevitably used for detecting, describing and matching fingerprint features such as minutiae and singular points. Using the ridge orientation, not only the error rates can be improved, but also more efficient image compression and a speed up in database queries can be achieved. This is the main motivation of this thesis and of many publications available in literature for modelling fingerprint ridge orientation. In this thesis we analyse current available techniques and propose a novel method for fingerprint ridge orientation modelling. One of the main problems it addresses is smoothing orientation data while preserving details in high curvature areas, especially around singular points. We show that singular points, which result in a discontinuous orientation field, can be modelled by the zero-poles of orthogonal polynomials. The models parameters are obtained in a fast two staged optimization procedure. Another contribution of this thesis is the application of a priori knowledge in fingerprint orientation models. Starting from the view point of flexible templates models, we develop a method which constraints the fingerprint orientation to vary only in ways as they occur in nature. Extensive experiments using a commercial state-of-the-art fingerprint matcher, have been carried out. We can report statistically significant improvements in both, singular point detection and matching rates.