Evaluation of fingerprint orientation field registration algorithms

Abstract

The majority of modern, fingerprint registration algorithms are based on the alignment of minutiae features. However, shortcomings of this approach are becoming apparent due to the difficulty of extracting minutiae from noisy or low quality images. This papers explores a novel approach to fingerprint registration based on orientation field alignment. One main advantage of this method is that orientation fields can be computed reliably for poor quality images, providing a robust feature for registration. Three orientation field alignment algorithms are presented, and their performance is evaluated using an FVC2002 dataset.