Biometric Systematic Uncertainty and the User

摘要

Often biometric technology and system evaluators are concerned with capturing accurate performance estimates that predict performance for target populations of interest. Occasionally, predictions, and perhaps adjustments, for subpopulations based on convenient characteristics such as sex or age are examined. Few experiments have been performed to explicitly study the interaction of the user and the biometric system - how a user's behavior, cognition, perception, and anthropometric qualities affect performance metrics and error estimates. This paper summarizes three studies performed in this emerging field of biometric usability. Each study focused on fingerprint recognition and the systematic uncertainty in system performance induced by different human factors. First, the effect of feedback and habituation on image quality is examined. Second, results from a study on the influence of the height of the surface upon which the fingerprint sensor is placed are presented. A summary of an experiment on the effect of instructional materials on the timing and error rates for a 10-print capture process closes the paper.