Fingerprint images enhancement in curvelet domain

Abstract

Due to variety of fingerprint images in quality, it is essential to perform a fingerprint enhancement stage before extracting minutiae. Since the performance of an automatic fingerprint authentication system depends on accuracy of extracted features, designing an efficient and accurate enhancement module is critical. In this paper we propose a new fingerprint enhancement method based on Gabor filter in Curvelet domain which can improve the clarity and continuity of ridge and valley structures. In proposed method first we apply Fast Discrete Curvelet Transform (FDCT) on query image. Then Gabor filter is employed on the coarse scale coefficients and a soft thresholding function is applied on the fine scale coefficients. Finally we reconstruct fingerprint image using those modified coefficients. Our primary experimental results on a small test set, which includes 21 fingerprint images, show the promising performance compare to Gabor-based and Wavelet-based methods. © 2008 Springer Berlin Heidelberg.