Authentication Using Non Minutiae Features Of Fingerprints

Abstract

Fingerprint based authentication is perhaps the oldest method of individual identification. newlineWith the advancement of the biometric technology, the development of an efficient newlineautomatic fingerprint identification system (AFIS) is feasible. Fingerprint based newlineauthentication approaches have been extensively investigated in the literature and are newlinebroadly classified into the minutiae-based and the image-based. Minutiae-based newlineapproaches have a wide acceptance and are the backbone of the current available newlinefingerprint recognition products. The matching involves essentially the searching of the newlinebest alignment of the minutiae sets of the input with the template fingerprints. Minutiae-based approaches suffer from various problems like missing true minutiae, false minutiae newlinetype, lack of sophisticated alignment method, poor image quality and complex input newlineconditions etc. Additionally, the minutiae-based approaches do not fully utilize the rich discriminatory information available in the fingerprints. On the other hand, the image-based methods exploit the more discriminatory information unlike the minutiae-based newlinemethods. Latest research on the image-based fingerprint matching approaches reveal that they have lower complexity than the minutiae-based approaches in dealing with the low newlinequality images. In this thesis, our aim is to devise non-minutiae features for the newlinefingerprint matching system for person authentication. newlineIn this thesis, we present: newline(i) A critical review of literature on the image-based fingerprint matching newlineapproaches. These are classified into Gabor-filter, transform and texture based newlineapproaches. The initial work on statistical texture based fingerprint matching newlinesystem is also included. newline(ii) A robust fingerprint matching method that exploits the orientation features newlineextracted from the circular region of interest (ROI) of radius 50 pixels centered at the reference point...