Abstract

    This paper presents an architecture and design for a multi-model human computer interface system used to control the mouse movements and execute mouse and windows commands using gaze tracking and speech recognition. The system is to help handicapped people, whom have severe disability in using their hands. The paper proposed optimum iris detection and tracking methodology integrated with an investigated method for gaze estimation and mapping. Our proposed method for gaze estimation and mapping is a hybrid method of the “2D Direct mapping” method and the “2D Reference mapping” method. Experimental results show that our system achieved 92% accuracy for iris detection and tracking. Our investigated 2D Gaze mapping method also achieved a high mapping accuracy and smoothing cursor pointer transition with the ability to recover when the user moves away from the camera and returns again. The method reduced the computational and calibration effort more than any previous used 2D mapping methods. We extend our system ability by adding a speech command recognition feature that enables the disabled user to execute other mouse and windows commands that are complex to recognize using the eyes