Gaze Estimation of Human Eye

As an indispensable component of intelligent transportation systems, the driving assistant has attracted much attention of scholars and engineers. The gaze estimation of human eye just serves for the driving assistant. In this paper the eye and gaze models are introduced first and then improved by us, so that the new models are presented. Based on the new models the Hough transform method with gradient direction is used to perform the iris detection and gaze estimation. In experiments the two parameters of false estimation rate and average detection time are investigated. To the good-quality frame images the false estimation rate is 5.87% and the average detection time is 94.3 milliseconds. In the end the proposed gaze estimation algorithm is applied to a real-time gaze tracking system and the comparison result shows that it has the more accuracy