M.Sc. Tezi Görüntüleme

       Eye gaze direction detection is the process of measuring the position of users gaze on computer screen. This gaze information is mostly used in the research areas like Human Computer Interaction (HCI), virtual reality applications, usability tests, eye typing for physically disabled people and drowsiness detection. In HCI, it is widely being employed as an alternative of traditional input devices such as mouse and keyboard.

       In this thesis, gaze image data and screen which user has interaction are mapped using appearance based video oculography approach and gazing directions of left, right, lower, upper, center and closed eye state are detected. Various feature extraction and classification methods are used on proposed eye models to improve the detection of gaze direction. Additionally, selection of the best approaches for gaze direction detection is performed by using the resulting classification accuracy of applied methods.

       In the proposed system, all components and methods are selected according to real time applicability and the resulting system applied on a few HCI problems as an alternative of user interface problems which gives successful results.

       Key Words: Eye Gaze Direction Detection, Human Computer Interaction, Appearance Based Video Oculography, Support Vector Machines, Local Binary Pattern Histograms, Eigen eye Approach, Haar Cascade Classifiers