Novel Approaches for Minutiae Verification in Fingerprint Images
Sharat Chikkerur, Sharath Pankanti, Nalini Ratha, Ruud Bolle, and Venu Govindaraju
Indian Conference on Vision, Graphics, and Image Processing, ICVGIP-04
(in submission)
Minutiae extraction is one of the critical steps in fingerprint verification algorithms. Any missing minutiae or spurious minutiae introduced at this stage can degrade the performance of the matching algorithm. Existing approaches for minutiae filtering use heuristics and adhoc rules to eliminate such false positives. We propose two new techniques for minutiae verification based on novel gray level features. We use directionally selective steerable wedge filters to differentiate between minutiae and non-minutiae neighborhoods with reasonable accuracy. We also show how the proposed feature can be used to build a minutia detector that operates directly on the gray scale images. We present a second technique based on Gabor expansion that results in even better discrimination.