Fingerprint image quality assessment, verification, and detection

Fingerprints are one of the oldest and most extensively used techniques for biometric verification/ identification of human beings. There are numerous distinctive methods to capture, process, and recognize an image of the user's finger. However, many real life fingerprinting problems such as a) fingerprint image quality assessment; b) fingerprint verification, which determines if two fingerprints were acquired from the same finger, with the highest possible reliability; c) detection, which is used to identify a given fingerprint from a database, still remain unsolved. Other challenges in fingerprint recognition include distortion of the fingerprint image due to dirt, cuts, scars, sweat, dry skin etc. Furthermore, the time required for verification/ identification increases non-linearly with the size of the database.

The main aim of this thesis is to develop a completely automated fingerprint processing system, which involves fingerprint image quality assessment, classification, and matching. A novel fingerprint quality assessment method is proposed to estimate the quality and validity of fingerprint images. Additionally, the complexity of matching is reduced by classifying fingerprint images using Gabor filter features. Finally, a robust fingerprint matcher, which combines both minutiae and local image features is proposed. In conclusion, the proposed automated fingerprint processing system contributes to the reduction of False Acceptance Rate and False Rejection Rate of the existing fingerprint matching systems.