Abstract. In this paper we utilise the minimum average correlation energy (MACE) and unconstrained minimum average correlation energy (UMACE) filters in conjunction with two correlation plane performance measures, max peak value and peak-to-sidelobe ratio, to illustrate the effectiveness of correlation filters in facial recognition login systems. A new technique for determining performance thresholds with the individual filters derived from the AMP and Biodm facial databases, was successfully developed producing high recall rates (67-94%) with 100% precision. A comparison of the precision and recall statistics obtained from the two different correlation plane measures, further demonstrated that max peak value is the better performance measure for use with MACE and UMACE filters for facial recognition login access control.
Daniel E. Riedel, Wanquan Liu, Ronny Tjahyadi