Fingerprints are one of the unique physiological characteristics of human beings. Fingerprints as biometric modality proved accrate, safe, and not complicated in the recognition process when compared with other modalities. This research is comparison study of the non-minutiae based to minutiae based fingerprint recognition software. As representation of non-minutiae based is self-developed software using wavelet methods and back propagation. Furthermore representation of minutiae-based is proprietary software called VeriFinger. The performance analysis considered in this research is their accuracy and processing time. The common problems faced in the fingerprint recognition are the quality of fingerprint image. The experiment was conducted using 750 fingerprint image with various qualities, and then there are 16 fingerprint images that do not pass VeriFinger threshold criteria. The accuracy of non-minutiae based software is 94.96%, while the minutiae based software is 98.37%. Average processing time of non-minutiae based software is 572 ms, and the minutiae based sofware is 374 ms. From the experiment, it can be concluded that the proprietary minutiae-based software VeriFinger is more accurate than the self based non-minutiae software. The reasoning of this conclusion come from the fact that minutiae based can extract the local characteristic patterns. But its faster in processing time cannot be judged well due to the VeriFinger code already have been optimized.
Gunawan A. A. S, Gazali W and Senjaya B. (2012). Analisis Sistem Indentifkasi Sidik jari Berbasis Minutiae dan Non_Minutiae. Proceedings SEMANTICS 2012, 5-9. BINUS Uiversity, Jakarta: Computer Science Department, School of Computer Science, BINUS University
Pattern Recognition, Fingerprint, Non-Minutiae, Wavelet Transform, Minutiae, VeriFinger