Title :
Hierarchically linked extended features in fingerprints
Author :
Krzysztof Mieloch;Axel Munk;Preda Mihailescu
Author_Institution :
University of G?ttingen, Institute for Mathematical Stochastics, Germany
Abstract :
Requirements for identifying and defining additional fingerprint features beyond minutiae, even not limited to level 3 details, has already been addressed by the community . However, new feature proposals are based on refinements of existing features (e.g. a finer level of classification), or on the introduction of new single features (e.g. 3-D level features such as the ridge height) . The new set of features proposed in this work does not only include additional fingerprint features individually but it also contains the information about their relationships such as line adjacency information at minutiae points or links between neighboring fingerprint lines. The dual information, that is the interference between ridges and valleys, is included as well. For the first confirmation of the quality of extracted features in application for the matching, the goodness of the extracted minutiae has been tested. The experiments have shown a decreased error rate when applying the minutiae to a standard minutiae-based matcher.
Keywords :
"Feature extraction","Fingerprint recognition","Bifurcation","Data mining","Distance measurement","Noise","Shape"
Conference_Titel :
Biometrics Symposium, 2008. BSYM ´08
Print_ISBN :
978-1-4244-2566-2
DOI :
10.1109/BSYM.2008.4655522