Eye Center Detection Using Combined Viola-Jones and Neural Network Algorithms

In eye tracking, calibration plays as a first crucial process towards detecting the eye. This process required the user's information to learn and maximize the accuracy of the eye tracking. Such that, various techniques can be used to detect and train eye gaze parameters. Thus, as the first step, the calibration points have been set on the user's screen display for the eye tracking process. The process of calibration is done by tracking the user's attention at 5 and 9 points of calibration process using a webcam. The circle detection of iris is used to determine the eye center coordinates. In this paper, the eye center coordinates are being extracted by using the well-known face detection algorithm which is Viola-Jones algorithm. Then, the coordinates will be trained as a data set for gaze location by using neural network algorithm. Results have shown that eye center can be well detected by using combination of both algorithms.