Fingerprint minutia recognition with fuzzy neural network

Abstract

Falcon-ART is a fuzzy neural network that can be used as fuzzy controllers or applied to areas such as pattern recognition, forgery detection and data analysis. Our previously proposed Falcon-DIC has stronger noise tolerance capability compared to the original Falcon-ART by employing a new clustering technique called Discrete Incremental Clustering (DIC). In this paper, Falcon-DIC is applied to perform direct gray-scale minutiae extraction. Fingerprint features extraction, or minutiae extraction, is an essential part of fingerprint identification systems. Most existing minutiae extraction methods require image preprocessing, such as binarization and thinning. Since these image processing techniques results in the loss of valuable information, our proposed approach can extract minutia directly from gray-scale fingerprint images. Experimental results show that Falcon-DIC based minutiae extraction has invariant ability to rotation and good performance on true acceptance. © Springer-Verlag Berlin Heidelberg 2005.