Touchless Fingerprint Matching Using FingerCodes in Comparison to Minutia-based Approach

This paper proposes FingerCodes as a touchless fingerprint matcher and compares it with the traditional minutia- based approach. First, the input images are processed with a touchless-to-touch algorithm in order to generate touchbased equivalents used in the minutia-based matching evaluation. Ragarding FingerCodes, each touchbased equivalent is used to estimate a reference point and a region of interest is defined. Pixels belonging to the original touchless image that fall within this region are divided into sub-blocks, normalized, filtered with a bank of Gabor filters, and FingerCodes are finally extracted. In our experiments, 400 touchless fingerprints are used. Matchings are performed using bootstrapping and the observed average equal error rates are 9.32% and 4.29% for FingerCodes and for minutia-based matching, respectively.