| In this thesis, a new technique for generating synthetic fingerprints is presented. Synthetic fingerprints are necessary in the following applications: (1) improving testing of fingerprint systems; (2) enhancing the security of a database of fingerprints; (3) training biometric systems.; The presented approach is based on a combination of the synthetic orientation map of fingerprints and a modified Gabor filter. The presented algorithm derives the orientation map of an original fingerprint image, then generates new orientation maps using the polar transformation, and synthesizes fingerprints using new orientation maps and a modified Gabor filter. The synthetic fingerprint images generated by the presented approach are similar to real fingerprints scanned from sensors with good quality. They can also be used to test biometric systems by adjusting control parameters flexibly, in particular, level of noise.; The project was implemented in the Biometric Technologies Laboratory, Electrical and Computer Engineering Department, University of Calgary. A package for the generation of synthetic fingerprint images was developed using Matlab. The results were compared with those generated by SFinGe, a commercial generator. The quality of the synthetic images was evaluated using the Comnetix fingerprint recognition software (Comnetix Inc.) available in the Biometric Technologies Laboratory. |
