Fingerprint verification using filter-bank of Gabor and Log Gabor filters

Abstract

This paper describes efficiency of Log Gabor filters in fingerprint verification by giving a performance-based comparative analysis with Gabor filter-bank based approach. The proposed method locates the core point through complex filtering technique and then tessellates the image around core point. A bank of filters is used to capture requisite distinctive fingerprint features as a fingerCode and matching is based on Euclidean distance between FingerCodes of query and registered images. A bank of six Log Gabor filters outperforms Gabor filter approach in terms of acceptance and success rates achieving 97% GAR, 3% FRR and 1% FAR whereas Gabor approach pulls off similar GAR but FRR and FAR swell up to 3.1% with a preset dataset of fingerprint images. It has also been noticed that the images, falsely accepted or rejected in case of Gabor filter-bank approach, were correctly verified through Log Gabor method.