Gradient based fingerprint verification using principal components

Abstract

This study present a method for fingerprint recognition based on principal component analysis (PCA) and point patterns (minutae) obtained from the directional histograms of a fingerprint. We first enhance the fingerprint images, then employ Principal Component Analysis (PCA) method to fingerprint data. The compressed data is then used for directional image representation. After the compressed data are obtained, the process continues with directional image formation, directional image block representation, and fingerprint matching, respectively. Our method determines the direction of each pixel, process the images in blocks and uses directional histograms, thus removes the need for thinning. The method gives same performance as that of the uncompressed data, but reduces computation. Furthermore, the parts of the system that successfully use artificial neural networks (ANN) are mentioned. © 2008 Elsevier B.V., All rights reserved.