Implementation and Evaluation of NIST Biometric Image Software for Fingerprint Recognition

Abstract: Fingerprints are rich in details which are in the form of discontinuities in ridges known as minutiae and are unique for each person. The main desire of this thesis is to implement and evaluate National Institute of Science and Technology (NIST) Biometric Image Software (NBIS) for fingerprint recognition using C language MEX-files. NBSI source code is written in ANSI C programming language. A system developed in Matlab environment using C language MEX-files is used for implemented and evaluated the NIST source code. Minutiae extraction and minutiae matching are performed by using a function developed with NBIS. The extracted minutiae are passed through the function with NBIS that match two minutiae patterns and produce a match score. Finally the performance of the system is evaluated on various fingerprint databases and results are presented in Receiver Operating characteristics (ROC) graphs.