Liveness Detection and Recognition System for Fingerprint Images

Abstract

In recent years, for verification and identification, uses of biometrics information are increased rapidly because they are more secure and reliable. Biometrics (face, finger, iris, palm, etc.) recognize the person based on human traits that are physiological and behavioral traits. Fingerprint recognition systems are widely used biometric for verification and identification due to its universality in nature and easiness. But they are various types of attacks are present that affect the performance of the fingerprint recognition system like spoofing attacks, displacement error, and physical distortion, etc. In this proposed system, work is carried out to overcome these types of errors and enhances the accuracy of the system. For spoofing detection, supervised learning with minutiae extraction method is used, for displacement error, alternating direction method multiplier (ADMM) is used and enhance the accuracy of the system by a technique that uses crossing number for minutiae extraction, for feature extraction gray-level difference method, discrete wavelet transforms, and feature matching using hamming distance. For learning and classification, support vector machine is used. In this fingerprint verification competition (FVC) 2002, FVC2004, FVC2006, and ATVS are considered for testing purpose and calculation of accuracy.