Research On The Technology Of Eye Tracking

Eye tracking technology is deemed as the most effective means in the research on visual information processing, and it has attracted the common concern of psychology, ergonomics, computer science and other fields. The results of it have been widely applied in industrial, commercial and military fields.The theoretical foundation of the eye tracking technology is discussed in the paper first. Then, on the basis of analyzing a variety of current technologies, combined with practical guidelines, eye tracking technology based on pupil-corneal vector method is deeply researched in the paper. Through the detailed research on the current algorithms which are used to determine the pupil center, the method that determines the pupil center by extending radials is adopted. The results show that the data that should be processed can be significantly reduced, and the robustness is improved when the initial point is poor. The subsequent simulation and analyses indicate that the algorithm can converge rapidly, and the inliers detected are dominant in the number.Due to the high demands of the eye tracking technology on time and accuracy, through the analysis of the limitations of the least-square method, considering the random sample consensus principle and the character of the characteristic points detection algorithm, a method that is based on random sample consensus is proposed to determine the pupil center, the biggest merit of which is that the operation speed is improved and the accuracy is maintained. The simulation results show that when the eyes look forward, the relative error of the pupil center is 0.53%, When the eyes look down, the relative error is 0.76%, and the time that this algorithm consumes is only 0.3203 that the RANSAC does. Furthermore, the determination of pupil center when there is shelter is also considered in the paper, and the simulation results validate the performance of the algorithm.