Improving Human-Computer Interaction by gaze tracking

We present a novel calibration method that enables tracking motion of user´s eye and gaze by using a single webcam. Human-Computer Interaction (HCI) is significantly influenced by the communication tool used by a human to control the computer. Substituting some functions of the keyboard or the mouse with human gaze driven control opens a new depth of HCI. Human gaze is much more informative than moving the mouse, that is, not only the direction of the gaze, but also emotional factors can be detected. It highly increases the efficiency and comfort of a user while interacting with the computer. In this paper we discuss the difficulties of a gaze tracking system and give a novel solution to this problem in a simplified, special case, when the face is still while the eyes move. If this condition does not hold, we assume that precise face position and orientation are provided by a face tracker, such as CLM. Tests are presented on various datasets, while accuracy and limitations of our system are also analysed.