A Fingerprint Matching Algorithm for Programmable Smart Cards
Tommaso Cucinotta, Riccardo Brigo, Natale Marco Di · 2005
This paper presents a hybrid fingerprint matching algorithm for user authentication based on the fusion of heterogeneous schemes, and designed to run on programmable smart cards. The approach is based on the well known texture vector and minutiae based techniques, where image processing and feature extraction occur on the host, while the card device performs the final match against the onboard template, which is never revealed to the outside world. This increases the security of the template itself and of the applications using it. The matching algorithms have been tuned in order to achieve an acceptable performance despite computation and memory constraints. Experimental results, gath- ered from our implementation on a Java Card device, highlight the feasibility of the hybrid approach. Fur- thermore, they show to what extent it is possible to trade precision for speed in the verification process, using appropriate tuning of the on board matching parameters.