Topological model of fingerprint image

The paper describes a new fingerprint image model which consists of topological and geometrical features of minutiae neighborhoods. To create a model, the authors suggest to calculate the topological features in the neighborhood of each fingerprint minutiae: ridge ending or bifurcation. The topological feature is the interrelation between two minutiae. An example of topological feature is the fact that two minutiae are situated on one ridge. Then, the list of topological features is constructed. Description of each feature is extended with metric information: distances between minutiae. Further, the proposed model is used for fingerprint identification. The experiments have shown that simultaneous usage of topological and metric features significantly improves the accuracy of fingerprint identification. On public available databases FVC2004, at FMR\;=\;10$^{-3}$, FNMR is 2%.