Touch-less fingerphoto feature extraction, analysis and matching using monogenic wavelets

Despite the wide use and success of touch-based fingerprint identification, elastic deformation is one of the primary reasons for false non-matching in most of the minutiae based automatic fingerprint identification systems. The main cause behind the large variation of the intra-class matching score is due to the non-uniform pressure applied during acquisition of fingerprint image. To avail deformation-free acquisition and other advantages of hygiene and user convenience, touch-less 2D fingerprint (fingerphoto) matching systems have been gaining popularity. The already existing fingerprint matchers are not suitable for extracting reliable minutiae features from fingerphoto images due to varying illumination, contrast, and magnification which occur while capturing the fingerphoto. In order to make fingerphoto compatible for such matchers, effective fingerphoto enhancement is crucial. In this paper, we propose a fingerphoto based recognition system which comprises a novel monogenic wavelet based fingerphoto enhancement technique that uses phase congruency features. For the performance verification of our system, we created a fingerphoto database using a smartphone camera in an unconstrained environmental condition along with its corresponding live-scan images. The experimental results on this database, as well as on other standard fingerphoto databases show a significant improvement in Equal Error Rates (EER) achieved using the proposed system.