An efficient visual tracking method based on single CCD camera

Abstract

Most of the existing visual tracking methods require users to wear some related equipments or fix heads. It is not convenient to use and the accuracy is not high. The method proposed in this paper is based on digital image processing technology and geometry transform. The trouble of wearing related equipments is avoided. At the same time, this method can get higher tracking accuracy and meet the requirement of instantaneity. The 3D space problem will be switched to a 2D image when the real-time human eye image is captured. Pupil-corneal reflection principle is utilized to get auxiliary parameters. A quick eye positioning method based on gray level is suggested in this paper. A cycle elliptic fitting method used is suggested to locate the pupil center. The existing space mapping method based on projective geometry principle is simplified. The experiment was done in the lab. Four accurate crass-ratio parameters were achieved, and the gazing point estimation got relatively high precision. © 2012 IEEE.