Anisotropic Diffusion-Based Contrast Enhancement of Frontal Facial Thermograms to Extract the Facial Vasculature

Facial vasculature is the structure of arteries and veins in the face and finds applications in biometrics, animations, and cognitive and affective psychology. Thermal imaging is the only feasible non-contact way to obtain the structure. The blood flow in the facial blood vessels brings about a spatial surface temperature distribution. However, the pixel intensity variation in the captured facial thermogram is significantly low due to the absence of abrupt temperature changes in the face region. The traditional contrast enhancement methods do not highlight the blood-flowing zones, depicting the vasculature. Therefore, the issue of extracting the vasculature region by contrast enhancement is still a challenge in the thermal image processing community. In this work, we propose a contrast enhancement technique for facial thermograms by incorporating the philosophy of anisotropic diffusion. We name our proposed method as anisotropic diffusion-based contrast enhancement. The results show that the proposed method improves vasculature extraction using state-of-the-art techniques.