Eye-ball tracking system for motor-free control of mouse pointer

Recent trends and evolution in the domain of image processing have yielded to us several high quality feature detection algorithms. While the research in finding new algorithms is ever so rampant, there is a need for equal number of applications of such algorithms in order to realize their full potential and use for common man. This paper makes use of a mixture of Viola-Jones, Kanade-Lucas-Tomasi (KLT) and Circular Hough transform algorithms for developing a robust eye ball tacking system for controlling the mouse pointer. The added feature in the system is representing clicks. One blink represents one click and two blinks in a short time span of time represent a double click. The paper also describes some other algorithms which were used but failed in tracking features. Since the computer dependence has increased so drastically in recent times this system proves to be of a direct advantage to people with motor disorders to efficiently browse through their files on the computer. Different algorithms have each a different forte. So instead of making one algorithm very complex to perform well on all parameters, making the best use of all these three algorithms really simplifies the task and yields a better performance than the individual three. The system was tested in different lighting conditions and distances from the screen and yielded an accuracy of close to 96% in efficiently tracking the iris which is very good considering that this is a real time implementation. The ultimate goal of the authors is to build a software package out of this system and distribute it open source; so simplicity in implementation has been the imperative concern to invoke a better user understanding of the algorithm.