"Committing yourself is a way to find out who you are.

A man finds his identity by identifying."

-- Anonymous

 

Fingerprint Matching Using Level 3 Features

Fingerprint friction ridge details are generally described in a hierarchical order at three different levels, namely, Level 1 (pattern), Level 2 (minutiae points) and Level 3 (pores and ridge shape). Although latent print examiners frequently take advantage of Level 3 features to assist in identification, Automated Fingerprint Identification Systems (AFIS) currently rely only on Level 1 and Level 2 features. With the advances in fingerprint sensing technology, a systematic study to determine how much performance gain one can achieve by introducing Level 3 features in AFIS is highly desired.

Quality-based Fusion in Multi-biometric Systems

Dynamically assigning weights to the outputs of
individual matchers based on the quality of the samples presented at the input of the matchers can improve the overall recognition performance of a multibiometric system. We propose a likelihood ratio-based fusion scheme that takes into account the quality of the biometric samples while combining the match scores provided by the matchers.

3D Touchless Fingerprints

An on-going collaborative research project with TBS.

Fingerprint Deformation Esimation

Non-linear deformation causes fingerprint features such as minutiae points and ridge curves to be distorted in a complex manner. We propose a method to estimate the non-linear deformation using Thin-Plate Splines (TPS) based on minutiae correspondences and construct a user specific deformation model, which is then utilized for distortion correction of the template prior to matching.

Fingerprint Spoof Esimation

Non-linear deformation can be a natural solution for spoof detection since human skin has special elasticity property that artifical materials lack. The marjor advantage of deformation-based approach over pore (perspiration)-based approach is that neither higher resolution fingerprint readers nor longer duration of image acqusision is required. Our preliminary results show that the non-linear deformation made by a live finger is very different from that of a "gummy" finger. We also show that skin deformation is mostly concentrated in the slip region and is more non-uniformly distributed than "gummy" deformation.

Fingerprint Image Quality

Image quality has one of the largest impacts on biometric system performance. Poor quality fingerprints often result in a large number of spurious minutiae extracted. Two image quality measures, one global and one local, are developed. The global quality measures the energy concentration in certain frequency bands of fingerprints. The local quality caculates the coherence measure and gives a quality map of the given image. Both quality measures are shown effective in predicting system performance at all processing stages, namely, enhancement, feature extraction and matching.

Iris Image Segmentation

Iris region is defined as the annulus band between the pupil and sclera. In order to obtain the iris code for matching, a segmentation algorithm is developed to extract both inner(outer) boundaries of the iris and upper(lower) eyelids that may occlude the iris region. Our algorithm is highly accurate, compared to other algorithms quoted in literature.

Iris Image Quality

Iris features are very localized, especially, there are more texture in the pupil area than in the ciliary area. We propose a local quality measure of iris images based on 2D wavelets. The quality measure is shown to be very effective in predicting iris matching performance.