The Segmentation of the Supraorbital Vessels in Thermal Imagery

Thermal imaging techniques have been applied to detect and measure mental stress in polygraph screening and other applications. Mental stress is highly correlated with the activation of the corrugator muscle on the forehead. The vessels that supply blood to the corrugator muscle, proportionally to its degree of activation, are the supraorbital vessels. The rate of blood flow in these vessels can be indirectly measured via the intensity of heat emission from their segments. However, segmenting the thermal imprints of the supraorbital vessels is challenging because (1) they are fuzzy due to thermal diffusion, and (2) exhibit significant inter-individual and intra-individual variation. In this paper, a new segmentation method is proposed to extract the supraorbital vessels in thermal imagery. The new method features three steps: (1) automatic initialization of vessels; (2) automatic localization of the central lines of vessels; and (3) fast determination of vessel boundaries. The results show that the new method achieves high quality segmentation in both a simulated and a real dataset. The proposed method is expected to further increase the accuracy of stress measurements via thermal imaging.