Discrete eye tracking based on PCA and neural classifier

Abstract

this work presents the discrete eye tracking system which is based on principal component analysis (PCA) and the neural classifier. The system use head mounted device. The user wears an ordinary cap with mounted on it Web-cam and mirror which reflects the eye to the camera. The system was tested with two on-line applications. First - the control of the mobile robot in the maze and second application is the text writing program ldquoEyeWriterrdquo. Over 20 commands per second can be generated using presented eye tracking system.