Gaze Tracking System Based On LPM Feature And Ensemble SVR

Multimedia human computer interaction is a kind of integration technology of multimedia and human computer interaction, it is majoring in varieties of expressing information and how to interact with computer by ways which connect with multiple input and output facilities. As a rising technique, eye tracking and gaze input have already become a focus in the field of human computer interaction lately. And in this paper we also will do research in human eye tracking theories and human computer interaction systems based on gaze input and some related tasks.First, we begin with the current development of human computer interaction technology and computer vision theories, and the basic principles and methods of eye tracking technology are also discussed in brief. While concluding the development and research of them at the time, the research background and content of this paper are explained after that.Second, with the theory analysis of the initiative infrared radiation source based on corneal reflex-pupil Center (PCCR), a gaze tracking system based on the human eye "characteristics" is proposed. Specifically, the combined features of space coordinates and the LPM are fed into a kind of ensemble SVR regressor to match the gaze mapping function, in order to realize the human and computer interaction in the process of natural head movement in case of gaze tracking.The gaze tracking algorithm of this paper combines the space coordinates of the eye with the LPM characteristics of the eye as the gaze direction features, uses the ensemble support vector regression algorithm to predict the mapping function between the gaze direction and screen coordinates. Traditional infrared gaze tracking method usually takes pupil-glint vector, but pupil fuzzy borders, shape changes and other factors make the pupil center migration, and in addition, the reflection point sometimes appears to be too large that offsets the actual position, which caused the pupil-glint vector inaccurate and affect experimental results. Using LPM features of eye images, not only the texture changes when gaze direction change can represent, but also includes the pupil-glint vector information. This method can avoid the disadvantage of the calculation of pupil-glint vector, and can estimate the gaze direction more accurately.