Robust face region discrimination and eye tracking to the environmental changes

Eye-tracking technology has been studied in the field of Human-Computer Interaction as well as industrial applications such as advertisement, medical diagnostics and surgery. However, because the eye tracking is mainly affected by environmental changes like illumination, scale and image rotation, we do not only use Haar classifier to detect face region and eye area, but also take advantage of the Circular Hough Transform to extract the pupil from detected eye region. The detected pupil is tracked in real time by making use of the template matching. We confirm the robustness and usefulness of proposed algorithm through lots of real experiments in changing circumstances of illumination, scale and rotation.