The performance of any automatic fingerprint identification system AFIS depends highly on the acquired fingerprint image quality, which is mostly exposed to degradation and noise during the acquisition process or because of fingerprint sweat, smut etc. Most AFIS use some form of image enhancement in order to improve the recognizer performance. This paper presents a simple modification to the enhancement method based on Gabor filtering algorithm in order to improve image quality. It has been noticed that the computations of the local ridge direction and also the enhancement is affected heavily by the block size used therefore, an attempt has been made to develop a more accurate approach which is dynamically decides the block size. It is based on the local ridge frequency. The proposed enhancement algorithm has been tested to fingerprint images from FVC2002 and it shows a promising result |
