Date

2016

Author

Baltacı, Serdar

Metadata

Show full item record

Item Usage Stats

205
views

158
downloads

Cite This

Stress has several negative physiological/physical impacts in our lives. Hence, it is important to recognize stress during daily activities. The relationship between stress and physiological or physical signals has been studied for a long time. The aim of this dissertation is to detect stress remotely using pupil diameter and facial temperature analysis. For this purpose, we developed a stress triggering experiment in order to generate physiological/physical effects. Our experiment consists of 2 parts which are applied consecutively. The first part was used as a baseline for neutral emotion, in which neutral pictures of International Affective Picture System (IAPS) were utilized. In the second part, to generate stress, negative pictures of IAPS were used. To detect emotional state of the participants, pupillary and facial thermal responses were measured using a TOBII TX300 eye tracker and a FLIR SC620 thermal camera. Entropy in a sliding window was used to accommodate the time differences in the physiological rise and fall profiles of pupil and thermal data. Pupil and thermal features derived from the measured signals and the entropy based values were fused at the feature level. Finally, classification accuracy of stress was enhanced with machine learning techniques. We were able to identify stressful responses from the participants with an accuracy of 83.8% using AdaBoost and Bagging classification methods. Results also show that the experimental protocol we suggested for stress detection is highly applicable based on pupil diameter and facial temperature.   

Subject Keywords

Eye tracking., Facial expression., Stress (Psychology)., Stress (Physiology)., Imaging systems., Infrared imaging.

URI

http://etd.lib.metu.edu.tr/upload/12620630/index.pdf
https://hdl.handle.net/11511/26175

Collections

Graduate School of Informatics, Thesis