This paper summarizes a study carried out on data from the Face Recognition Vendor Test 2006 (FRVT 2006). The finding of greatest practical importance is the discovery of a strong connection between a relatively simple measure of image quality and performance of state-of-the-art ven- dor algorithms in FRVT 2006. The image quality measure quantifies edge density and likely relates to focus. This ef- fect is part of a larger four-way interaction observed be- tween edge density, face size and whether images are ac- quired indoors our outdoors. This finding illustrates the broader potential for statistical modeling of empirical data to play an important role in finding and codifying biometric quality measures.
Proceedings Title: 8th IEEE International Conference on Automatic Face and Gesture Recognition
Conference Dates: September 17-19, 2008
Conference Location: Amsterdam, NL
Pub Type: Conferences
Face Recognition, Performance Analysis, Generalized Linear Mixed Models