Real-time wide-view eye tracking based on resolving the spatial depth

Author：Haifeng Bao,Weining Fang,Beiyuan Guo,Jianxin Wang

Current eye-tracking systems include many types that have been applied in numerous applications; however, they are unsuitable for industrial environments such as aircraft cockpits and train cabs. While wearable eye-tracking glasses have a wide field of view, they provide only a relative point of view in a captured video image; thus, they cannot function as a real-time interface to digital monitors. Desktop eye-tracking applications can interact with a single screen in real-time using pre-attached fixed cameras or infrared sensors; however, these applications allow only a narrow field of view. In this study, we developed a novel eye-tracking solution that integrates both the requirement for interaction and a wide field of view. The eye-tracking glasses gather eye movement data while a motion capture device provides data concerning the position and orientation of the head. A spatial depth-resolving algorithm is proposed to estimate the distance from the eyes to the digital screen, making it possible to locate the screens.
Our proposed method is a generalized solution for estimating gaze points on wide screens; it is not dependent on specific devices. We tested the method in a virtual environment using unity3d and reached three conclusions: the algorithm theoretically has good accuracy and stability; it cannot be simplified; and when applied in a real environment it should have a satisfying and acceptable usability. Subsequently, we performed experiments in a real environment that validated the theory. Further applications:
This type of unique wearable equipment can function as a real-time machine input interface to enhance the machine’s perception of the user’s situational awareness and improve the machine’s dynamic service capabilities.

The research was supported by the National Natural Science Foundation of China under
Grant No. 51575037 and Research Foundation of State Key Laboratory of Rail Traffic Control and Safety under Grant No. RCS2018ZT009.

REFERENCES：Haifeng Bao,Weining Fang,Beiyuan Guo,Jianxin Wang,Real-time wide-view eye tracking based on resolving the spatial depth, MULTIMEDIA TOOLS AND APPLICATIONS,2018.11

https://link.springer.com/article/10.1007%2Fs11042-018-6754-2