Singular-minutiae points relationship-based approach to fingerprint matching

Abstract

Key issues with several of the existing fingerprint matching algorithms include problem of alignment of two minutiae feature vectors, failure with noisy image and nonlinear distortions. In this study, singular-minutiae points relationship-based algorithm is proposed for addressing the problem of alignment of two minutiae feature vectors in fingerprint matching. The algorithm uses the minutiae in the neighborhood of the core or delta point for promoting accuracy and reduction in computation by taking advantage of the fact that same source images of equal dimensions maintain same distance for every minutia point and the core/delta point irrespective of orientation. Results of experiments on local fingerprints and FVC2006 standard fingerprint databases were classified into correct, false positive and false negative. With correct results, the reference fingerprint is correctly matched to one or more fingerprints from the same person while in the case of false positive; the reference fingerprint is matched to one or more fingerprints of another person. False negative results were recorded for cases where the reference image refused to match with any of the fingerprints in the database. Based on FVC2006 fingerprints database, the Receiver Operating Characteristics (ROC) curves were also generated for the proposed algorithm and some recently formulated ones. Analysis of obtained results in all cases shows very good performance of the new algorithm.