ABSTRACT
The forensic fingerprint community has faced increasing criticism by scientific and legal commentators, challenging the validity and reliability of fingerprint evidence due to the lack of an empirical basis to assess the quality of the friction ridge impressions. This paper presents a method, developed as a stand-alone software application, DFIQI ("Defense Fingerprint Image Quality Index"), which measures the clarity of friction ridge features (locally) and evaluates the quality of impressions (globally) across three different scales: value, complexity, and difficulty. Performance was evaluated using a variety of datasets, including datasets designed to simulate casework and a dataset derived directly from casework under operational conditions. The results show performance characteristics that are consistent with experts' subjective determinations. This method provides fingerprint experts: (1) a more rigorous approach by providing an empirical foundation to support their subjective determinations from the Analysis phase of the examination methodology, (2) a framework for organizations to establish transparent, measurable, and demonstrable criteria for Value determinations, (3) and a means of flagging impressions that are vulnerable to erroneous outcomes or inconsistency between experts (e.g., higher complexity and difficulty), and (4) a method for quantitatively summarizing the overall quality of impressions for ensuring representative distributions for samples used in research designs, proficiency testing and error rate testing, and other applications by forensic science stakeholders.
Subject(s)
Dermatoglyphics , Software , Datasets as Topic , Humans , Models, StatisticalABSTRACT
The forensic fingerprint community has faced increasing amounts of criticism by scientific and legal commentators, challenging the validity and reliability of fingerprint evidence due to the lack of an empirically demonstrable basis to evaluate and report the strength of the evidence in a given case. This paper presents a method, developed as a stand-alone software application, FRStat, which provides a statistical assessment of the strength of fingerprint evidence. The performance was evaluated using a variety of mated and non-mated datasets. The results show strong performance characteristics, often with values supporting specificity rates greater than 99%. This method provides fingerprint experts the capability to demonstrate the validity and reliability of fingerprint evidence in a given case and report the findings in a more transparent and standardized fashion with clearly defined criteria for conclusions and known error rate information thereby responding to concerns raised by the scientific and legal communities.
