| Vision-based Human Computer Interaction (HCI) has the advantage of non-intrusiveness and naturalness. Human gaze, identifying a person's focus of attention, can provide useful communication cues to an HCI interface. The direction of a person's gaze is determined by two factors: the orientation of the face (face pose), and the orientation of the eyes (eye gaze). Head gesture, a kind of non-verbal interaction among people, also can serve as an interface for HCI. Basically, head gestures are interpreted from head movement, and face pose tracking is the first step for head gesture recognition. In this thesis, various computer vision algorithms are developed for eye tracking, gaze tracking, face tracking and face pose tracking in order to extract gaze information and head gesture information.; First, we introduce a new real time eye detection and tracking methodology that works under variable and realistic lighting conditions and various face orientations. Second, an accurate gaze estimation method is developed using the Generalized Regression Neural Networks (GRNN) to identify the mapping relationship from pupil parameters to screen coordinates. Third, based on the detected eye positions, we propose a framework for face and face pose tracking from a monocular image sequence.; All of these techniques are tested with subjects of different ethnic backgrounds, genders and ages, as well as subjects with/without glasses. Moreover, they are tested under different illumination conditions. Experimental study shows significant improvement of our techniques over the existing techniques. |
