Thermal Facial Signatures for State Assessment during Deception.

This study explored emotional and physiological states in response to stressful situations involving deception, and their relation to facial imaging. Male participants were evaluated (i.e., appraised) impending stressors. Stressor appraisals ranged from threat (i.e., appraisals that there are a lack of sufficient resources to meet the demands of a stressor) to challenge (i.e., appraisals that there are adequate coping resources to manage stressor demands). We used appraisals to discriminate two appraisal groups: threat versus challenge. The capacity to detect state changes of the human face was examined using several standoff sensing technologies, including 2D visible (VIS) and mid-wave infrared (MWIR) cameras. Psychophysiological determination of the human state was achieved through self-assessment and physiological measures which formed the ground truth for the classifier systems. Two deception studies, a false opinion study and a false behavior study, served as high-stakes stressors to understand facial changes with respect to human stress states. The methodology extracted MWIR statistical features from facial regions in response to changes in stress state. Using the statistical features from MWIR sensor and implementing Eigen analysis allowed classification of the threatened and challenged participants for the false behavior study with an accuracy of 85%.