Research On Eye Gaze Tracking Technology Based On Monocular Vision

Unmanned weapon system is an important direction of the development of weapon equipment. For a long period of time in the future, human in loop teleoperation is the main control of unmanned weapon. Unable to meet the demand of real-time battlefield. Aim by eye tracking is a new type of human-computer interaction, aiming by the eye gaze direction can liberate the soldiers’ hands, can greatly improve the operational efficiency of the soldiers handling the unmanned weapon system. The eye tracking discussed in this article is the key of this technology.The method in this article is based on the pupil center cornea reflection technique. We design the head- mounted gaze tracking system, capture eye images by camera. Then program algorithms under the LINUX system program with OPENCV. Images are processed by Gauss filtering, ROI region extraction and gray scale. Determine the adaptive threshold based on gray level histogram and binarization segmentation of gray images. Extract the pupil contour by Canny edge detection. Determine the center of the pupil coordinates combined with the total least squares ellipse fitting algorithm. Obtain the coordinates of the Purkinje center by centroid method. And calculate the 2D offset vector P-CR. Finally, establish the mapping relationship between the P-CR vector and the fixed point in the form of polynomial function. Realize the transformation between the eye information and the gaze point information to achieve the human eye tracking. The experimental results show that the accuracy of eye tracking is more than 90%, which lays the foundations for the technology of eye tracking aiming at the unmanned weapon.