Fingerprint Recognition: Enhancement, Feature Extraction and Automatic Evaluation of Algorithms

Presentata Da:, Francesco Turroni
The identification of people by measuring some traits of individual anatomy, physiology or other behavioral characteristics has led to a specific research area called biometric recognition. Several biometric technologies have been developed and successfully deployed: fingerprints, face, iris, palmprints, signature. Fingerprints are the biometric trait discussed in this Thesis because of their individuality and persistence properties, as well as cost and maturity of products. This Thesis is
more » ... ed on improving fingerprint recognition systems considering three important problems: fingerprint enhancement, fingerprint orientation extraction and automatic evaluation of fingerprint algorithms. Fingerprint enhancement and fingerprint orientation extraction can be considered as pre-processing steps to the main goal in fingerprint recognition: the automatic comparison with other fingerprints (matching). An effective extraction of salient fingerprint features depends on the quality of the input fingerprint. If the quality is good, the fingerprint flow is well evident and a reliable set of features can be extracted. If the fingerprint is very noisy, we are not able to detect robust information: a large number of spurious features are extracted and we miss several genuine features. Therefore, to achieve high recognition performance it is essential to incorporate in the system a fingerprint enhancement module able to improve the quality of noisy fingerprints, thus making the subsequent processing steps more reliable. The goal of fingerprint orientation extraction is to compute one of the most critical information in fingerprints, the local orientation: a feature denoting the direction of the ridge flow at discrete positions. A precise estimation of the orientation field would greatly simplify the estimation of other fingerprint features (singular points, minutiae) and improve the performance of a fingerprint recognition system. Although new developments and improvements in fingerprint recognition are continuously reported, it is often difficult to understand, from the scientific literature, which are the most effective and promising methods. In fact, scientific papers typically pro-i pose recognition systems that integrate many modules (enhancement, feature extraction, matching, post-processing, etc.) and therefore an automatic evaluation of fingerprint algorithms is needed to isolate the contributions that determine an actual progress in the state-of-the-art. After a summary of state-of-the-art in fingerprint recognition, a new fingerprint enhancement method, which is both iterative and contextual, is proposed. This approach detects high-quality regions in fingerprints, selectively applies contextual filtering and iteratively expands like wildfire toward low-quality ones. The method does not require any prior information on local orientations or frequencies. We assess the improvements given by this algorithm over both real and synthetic fingerprints using a state-of-the-art matcher. The fingerprint orientation extraction is improved following two directions. First, after the introduction of a new taxonomy of fingerprint orientation extraction methods, several variants of baseline methods (local and global) are implemented and, pointing out the role of pre-and post-processing, we show how to improve the extraction. Second, the introduction of a new hybrid orientation extraction method, which follows an adaptive scheme, allows to improve significantly the orientation extraction in noisy fingerprints. It exploits both the local information and the experience, represented by the knowledge of plausible fingerprint orientation structures, to compute the best orientation at discrete points.