University of Houston

Stress is an integral part of everyday life. Quantifying stress has important applications in medicine and psychophysiology. Although many methods have been developed to quantify human stress, very few are passive, contact-free, and highly automated.

This dissertation studies the use of the forehead channel in measuring mental stress. The sensing modality of choice is thermal imaging. Thermal emission from the supraorbital vessels on the forehead is intensified when mental load is increased. This is due to increased blood flow to the corrugator muscle, whose activation communicates the existence of ongoing mental processes.

A new segmentation tool, Open End Snake (OES), has been designed to delineate fuzzy vessels in thermal images. It has three features: (1) Automatic initialization of vessels; (2) Automatic localization of the central lines of vessels; and (3) Fast determination of vessel boundaries. The OES outperforms other vessel segmentation methods in comparative experiments conducted as part of the study.

The new measurement methodology has been applied to lie detection and the detection of office-place stress. For lie detection, we analyzed data from two experiments. We have achieved good prediction rates in both experiments. For the detection of office-place stress, the relevant experimental results also demonstrated the application potential of the method.

A number of open issues and problems have thoroughly been studied and possible solutions have been debated and documented.