Using Eye Gaze For The Interaction With Natural Environments

Abstract

A Machine Learning based methodology is developed for estimating the Eye Gaze direction for the interactions and communications with the surroundings. Face features detections and iris localizations are one of the most dominant topics nowadays in this scientific research field. A robust methodology is developed in this project by mainly depending on the displacements of the face features, relatively to each other, and the iris/pupils within the eye's regions. In this thesis, a literature review is firstly conducted on the currently existing methodologies used to detect the face features and the irises. Subsequently, a comprehensive model is proposed based on the displacements of six facial features. The finalized methodology is then tested using MATLAB tools and reached an accuracy of 91.35%.

Date of Award	May 2020
Original language	American English