Image Segmentation-Based Face Tracking on Thermal Images for Automatic Estimation of Psychophysiological States Using Facial Skin Temperature Distribution
Abstract: In human-machine system, human and machine need to recognize each other’s state with continuously, quantitatively and real-time property. Facial skin temperature could be measured with these properties by infrared thermography. The non-contact property is a great advantage in bioinstrumentation. Previous studies have been reported the availability of facial skin temperature for evaluation of psychophysiological states of a human such as stress, drowsiness and emotion. On the other hand, the development of the face detection and tracking techniques on thermal images are necessary for the automatic evaluation of psychophysiological states of a human based on facial skin temperature, measured by infrared thermography. The objective of this study is to establish the technique for face detection and tracking on thermal images. In this study, the algorithm consisting of three phases: (A) human detection based on inter-frame difference, (B) face detection based on image segmentation, and (C) face tracking based on temporal analysis, is proposed. As a result, the face region on thermal images could be detected and tracked with high precision. However, a part with low temperature such as the back of nasal and cheek was classified as a region other than a face.
Keywords: Thermal image; face tracking; image segmentation