EXPERIMENTAL EVALUATION
FOR BASELINE RESULTS
Toestablish the baseline performance on the MOL Fdatabase. several experiments are performed. These experiments are
designed to demonstrate the challenges associated with the proposed database and to highlight its usage. The baseline
results for livescan fingerprint experiments are computed using two fingerprint matching algorithms: NBIS
(NIST Biometric imaging Software) 1321and VeriFinger 136J. NBIS is an open source minutiae based matching algorithm developed by NIST whereas VeriFinger is a low cost proprietary software by Neurotechnology. Latent fingerprint matching is an open research problem at the community is attempting to address. It is important to note that there is no standard latent fingerprint matching Software Development Kit (SDK) or commercial system available in the public domain, using which baseline performance can be established. In literature, we have observed that local Minutiae Cylinder Code (MCC)137],1381 description for manually marked minutiae provides state-of-the-art results 139l. Therefore, MCC descriptors are utilized for establishing baseline results on the latent fingerprint dataset. First, a NFIQbased I401 analysis is performed to understand the quality distributions of different subsets of the databases. Thereafter, three different sets of experiments are performed to establish the baseline in different application scenarios. The first experiment (Experiment evaluates the performance of optical scanner fingerprints while the other two experiments (Experiment ll and Experiment pertain to latent fingerprint matching. For Experiment. both and verification experiments are performed and the results are reported using the Cumulative Match Characteristics (CMC) curve and the Receiver Operating Characteristics (ROC) curve. respectively. For Experiment I and Experiment ll, identification experiments are performed and the results are reported in terms of the CMC curve.