Performance optimization in fingerprint analysis

ivPERFORMANCE OPTIMIZATION IN FINGERPRINT ANALYSIS(Ph. D. Thesis)Salih GÖRGÜNOĞLUGAZI UNIVERSITYINSTITUTE OF SCIENCE AND TECHNOLOGYDecember 2006ABSTRACTToday, fingerprint has a wide usage in several areas. It is especially used inhuman identification and verification, which is more reliable than thetraditional methods used for Access, such as passwords, personal identificationnumbers and keys. In this study, the algorithms, used in fingerprintidentification and verification systems have been examined, by considering theireffect on the systems performance. Three algorithms, namely: fingerprintsegmentation using the histogram of threshold value derived from Otsuapproach, fingerprint enhancement by the employment of a directional mask,and fingerprint verification by correlation analysis have been developed. Anautomatic fingerprint identification and verification software has beendeveloped with the employment of C++ Builder visual programming language.Therefore, the affect of algorithms on the overall system performance have beenevaluated. The performance of the algorithms found in the literature have beencompared with the developed ones. It has been shown that, they contributetowards increasing the system performance.Science Code : 702.1.014Key Words : Fingerprint identification, Fingerprint enhancement,Fingerprint matching, Segmentation, Minutiae extractionPage Number : 113Adviser : Doç. Dr. Abdullah ÇAVUŞOĞLU