The Analysis and Application of Facial Sequence Images with an Infrared Thermal Approach

Abstract

In this paper different types of emotion images selected from the international affective picture system were presented to subjects during infrared thermal facial image approach in order to collect sequence images for further analysis and evaluation of changes in facial temperature First a set of facial infrared thermal sequence imaging procedures to reduce the image deviation caused by unconscious head shaking without compromising the subject's comfort and resulting in minimal damage is proposed A thermal image is positioned by the center of mass of the image binocular area to determine the image translation and rotation processing and to output a fixed image for calibration In this study the automatic calibration of thermal sequence images was completed by the proposed two-stage gene algorithm and the difference in images was calibrated by image quality evaluation Through the above process the precise face images were aligned the automatic calibration of the sequence image for facial area was segmented and then the mean temperature of each subject's forehead nose mouth left cheek and right cheek areas were calculated The features was dimensioned by the principal component analysis method the significance was analyzed by using the multivariate variance number and the support vector machine classifier was used to verify the temperature difference of the face of patients with moderate and severe levels of schizophrenia The results were consistent with the theory as the frontal lobe produced an asymmetrical emotional with an identification rate of up to 94 3% Finally we propose an algorithm to construct the map of emotional significance to disassemble the independent time series and the corresponding independent spatial composition Analysis of variance was used to calculate the maximum correlation between the emotion sequence template and the time series This method can visualize the significant evoked emotional and facial temperature changes

Date of Award	2017 Oct 24
Original language	English
Supervisor	Chieh-Li Chen (Supervisor)