Abstract
This paper presents an effective method to detect face in thermal imaging. It utilises the head curve geometry to extract the face from the images. Thermal imaging system is chosen due to its inherent nature of being immune towards drastic ambient light changes. A detailed face detection algorithm is explained in the paper, together with the performance comparison. Several conditions have to be fulfilled prior to the face detection. There should be one person facing the camera and no other external heat-emitting objects are captured by the camera. Overall, the face detection method demonstrates a high accuracy, which is 90.68% for the near images, 92.12% for the far images and 91.4% as an overall.