Affective state aware biometric recognition

This paper examines the influence of differentiations in emotional and physical states on the user recognition potential of activity-related biometric methods. In this respect, the efficient anticipation and modelling of the affective state of the average user is attempted herein, so as to provide valuable information to the recognition process in terms of both accuracy and robustness. Following the corresponding findings, a novel probabilistic framework is proposed for compensating the falsely rejection decisions by enhancing the recognition performance on the basis of affective and corporal indicators. As a starting point, the focus of this research is over the states of stress and fatigue, however the proposed framework can be easily extended so as to include further affective indicators, if uncorrelated with each other. The proposed framework has been experimentally validated on a dataset of 25 real subjects, while the final outcomes exhibit significant advances in the recognition performance.