Robust Fingerprint Recognition - Asker Bazen, University of Twente, Enschede, The Netherlands

Fingerprint recognition is one of the most widespread biometric technologies because of its ease of use and its relatively high recognition performance. There are many commercial vendors of fingerprint recognition technology, but at the same time academic research is still needed to improve the fingerprint recognition performance. This talk will give an introduction to fingerprint recognition methods, with a focus on algorithmic aspects. It starts with an analysis of the fingerprint-recognition problem, which results in an overview of the causes of recognition errors. Next, three different fingerprint matching algorithms are explained: (1) gray-scale pattern matching, which is a straightforward but computationally expensive method, that may in some cases be able to provide accurate recognition decisions, (2) elastic matching, which compensates for elastic deformations in fingerprint images, (3) shape matching, which can be used for efficient database search. Finally, an overview of open research topics will be given.