Influence of the datasets size on the stability of the LR in the lower region of the within source distribution

Rudolf Haraksim, Didier Meuwly

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

    123 Downloads (Pure)

    Abstract

    This article focuses on the statistical evaluation of the fingermark evidence using the likelihood ratio (LR) approach. It studies the influence of the quantity of data used to model the within (WS) and between (BS) source variability. The LR system built for the experiment uses an Automated Fingerprint Identification System (AFIS) feature extraction and comparison algorithm, fingermark and fingerprint datasets coupled with a generative approach for modeling the WS and BS variability. This article concentrates on the computation of LRs of the same source in the lower region of the WS distribution. It analyzes the behavior of the LR with an increasing number of entries in the WS datasets while maintaining the constant proportion of the BS dataset in an attempt to estimate the amount of same source scores necessary to achieve consistent LR performance.
    Original languageUndefined
    Title of host publicationProceedings of Biometric Technologies in Forensic Science, BTFS 2013
    Place of PublicationNijmegen
    PublisherRadboud University
    Pages34-38
    Number of pages5
    ISBN (Print)2351-9738
    Publication statusPublished - Oct 2013
    EventProceedings of Biometric Technologies in Forensic Science, BTFS 2013, Nijmegen: Proceedings of Biometric Technologies in Forensic Science, BTFS 2013 - Nijmegen
    Duration: 1 Oct 2013 → …

    Publication series

    NameBiometric Technologies in Forensic Science
    PublisherRadboud University Nijmegen
    ISSN (Print)2351-9738

    Conference

    ConferenceProceedings of Biometric Technologies in Forensic Science, BTFS 2013, Nijmegen
    CityNijmegen
    Period1/10/13 → …

    Keywords

    • EWI-25377
    • METIS-309696
    • IR-93344
    • Forensic Science

    Cite this