The field of biometric face recognition blends methods from computer science, engineering and statistics, however statistical reasoning has been applied predominantly in the design of recognition algorithms. Here we broadly review face recognition from a statistical viewpoint and direct attention to the growing field of bio- metric performance evaluation. Methods for performance evaluation seek to identify, compare and interpret how characteristics of subjects, the environment and images are associated with the performance of recogni- tion algorithms. Although the the design and evaluation of face recognition algorithms draw upon some famil- iar statistical ideas in multivariate statistics, dimension reduction, classification, clustering, binary response data, generalized linear models and random effects, the field also presents some unique features and chal- lenges. Opportunities abound for in-depth and innovative statistical work in this field. We review some central topics in face recognition and illustrate the evaluation problem by summarizing a generalized linear mixed model analysis of the pre-eminent face recognition dataset used to test state-of-the-art algorithms. Findings include that (i) between-subject variation is the dominant source of verification heterogeneity when algorithm performance is good, and (ii) many covariate effects on verification performance are universal across easy, medium and hard verification tasks.
Citation: Computational Statistics
Pub Type: Journals