Design and development of an automatic fingerprint verification system

Abstract

Data security is an important part of internetworking. It prevents fraudulent users from accessing an individual personal data. Biometrics is one such authentication method used in a wide range of application domains such as ecommerce and automated banking. Biometrics is more reliable and more capable of differentiating between an authorised person and a fraudulent impostor than traditional methods such as passwords and PIN numbers. There are a number of biometrics technologies being researched and under development such as fingerprint identification, face recognition, iris recognition, etc. However, fingerprint identification is one of the most reliable biometrics technologies. Generally, there are two approaches to fingerprint identification, namely conventional and bypass. In the former approach, fingerprint images have to go through several processes including noise removal, segmentation, thinning and finally minutiae extraction. Whereas, in the latter approach, the minutiae are directly extracted from a greyscale image and bypassing all the above processes. However, the minutiae extraction is an error prone process, depending on quality of the fingerprint images. A low quality image will generate many false minutiae that eventually lead to errors in fingerprint identification. This research focuses on design and development of an automatic fingerprint verification system that capable of handling a wide variety of fingerprints. Here, the biggest challenge is to develop a technique that can enhance or improve a low quality image which contains scars, sweat spots and broken ridges. Our proposed framework is started with noise removal, and followed by image enhancement, directional image computation, fingerprint reconstruction, segmentation, thinning, minutiae extraction, and finally fingerprint matching. In this study, 500 fingerprints were tested and the percentage of successful matches was 91 percent. This achievement was directly attributable to our new enhancement techniques excellent performance in the fingerprint reconstruction.