Design And Implement Of Eye Tracking System Using Intel Perceptual Computing

Eye tracking technology provides a more intuitive and natural way of human-computer interaction, which can bring new experience to the users. After years of exploration and research, eye tracking technology has made breakthrough progress. Especially, Intel recently launched Perceptual Computing technology, eye tracking as an important research direction of Perceptual Computing, has a broad application prospect. In the near future, Perceptual Computing will be integrated into the Intel architecture equipment. As a result, the eye tracking system based on Intel Perceptual Computing technology will be an important development and application direction in the field of human-computer interaction.This thesis firstly introduces the background and significance of eye tracking technology, reviews the research of eye tracking technology in recent years. And the basis of the theory and technology that the eye tracking system development requires are also discussed. Additionally, on the basis of Intel Perceptual Computing technology, an eye tracking system is built, the overall architecture of the system and the detailed implementation process of the system are given.In the eye and pupil detection module, the thesis proposes a pupil detection method based on connected region labeling, it employs Intel Perceptual Computing SDK to track the eye, and then two-pass connected region labeling algorithm is utilized to achieve the detection of pupil. By using this method, the average pupil detection accuracy is up to 88% and the frame rate reaches more than 20 FPS. For gaze mapping and evaluation module, a simple, rapid and efficient gaze mapping and evaluation method is studied and proposed. By constructing the theoretical model to quickly generate calibration matrix and collecting adequate pupil data to generate calibration rectangle, the pupil points in the calibration rectangle are normalized to the points in the model rectangle, then the points in the model rectangle are mapped to the screen region according to the calibration matrix, so as to realize the mapping of the pupil point. Using this method, the gaze tracking accuracy is 86% on average and frame rate is up to 20 FPS. Finally, the eye tracking system API is provided, and on the basis of the API, the thesis develops and implements an application program:scrolling page of the browser. Actually, it shows that the eye tracking system using Intel Perceptual Computing technology can basically meet the application requirements.