Performance of Biometric Quality Measures

Abstract

We document methods for the quantitative evaluation of systems that produce a scalar summary of a biometric sample's quality. We are motivated by a need to test claims that quality measures are predictive of matching performance. We regard a quality measurement algorithm as a black box that converts an input sample to an output scalar. We evaluate it by quantifying the association between those values and observed matching results. We advance detection error trade-off and error versus reject characteristics as metrics for the comparative evaluation of sample quality measurement algorithms. We proceed this with a definition of sample quality, a description of the operational use of quality measures. We emphasize the performance goal by including a procedure for annotating the samples of a reference corpus with quality values derived from empirical recognition scores.

User identity verification via mouse dynamics

“…Results showed superiority of the action-based method proposed in this paper over the histogram-based method proposed in [1]. Furthermore, evaluation showed a significant improvement in the verification accuracy when using the newly introduced features.…”

“…Equal error rates (ERRs) of 0.007 and 0.002 were achieved for 100 and 200 mouse-strokes, respectively. Ahmed et al [1] monitored the mouse activity of users while they performed their daily tasks within their own chosen operating conditions and applications. Features were extracted and aggregated into histograms that were used to characterize each user.…”

“…Since biometric-based verification systems are a special case of classifiers [1], their performance is evaluated using similar measurements. Specifically, the following measurements were used:…”

“…The discretisized features were used by both the proposed approach and the histogrambased one. By performing aggregation of the discretisized features of each action, occurrence histograms as in [1] were created. The feature average histograms were created by averaging the remaining features.…”

A scheme for predicting recognition performance by using confidence intervals

“…P. Grother et al [1] reported that quality measures predict the authentication performance, and Y. Chen et al [2] proposed two such quality measures. K. Kryszczuk et al [3] reported a method in which classifier decisions and the corresponding reliability information are combined to predict and correct verification decisions.…”