Research on video control algorithm based on eye movement information

Abstract:
In this paper, there is no infrared light source in the line of sight, no eye tracker is used, and a relatively common and inexpensive camera is used to obtain the user's image, so as to realize the real-time automatic control of the video by the line of sight (the user does not need to make additional actions), which can be used to judge the user's perception of the video, and can also be used for contactless human-computer interaction technology to reduce the spread of COVID-19. This paper proposes a method for actively manipulating videos by analyzing the gaze direction of the human eye. The system collects user images through a web camera, and uses the 68 feature point model of the face provided by the dlib library to detect and extract the feature points. According to the circular aggregation and gray distribution characteristics of the pupil, the gradient consistency method is used to determine the position of the center of the pupil, and finally according to the pupil position, it is judged whether the line of sight falls on the screen, and then the video is operated accordingly. In this paper, a method for initial determination of orientation based on nose tip and pupil and a real-time video active switching algorithm based on spatial geometry and user eye movement information are proposed. The experimental results show that the interactive system can better operate the video according to the human eye movement information, and realize the convenient and concise human-computer interaction.