Coordination Eye Tracking Technology And Its Interactive Application

Eye tracking can record eye movement data,such as user's fixation coordinates,and,analyse user's visual behavior.Eye tracking technique can be widely used in a number of areas,such as reading behavior research,advertisement assessment,and human factors.With the development of digital image processing technieuqes,and computer supported cooperative work,eye tracking has also begun to apply to collaborative interaction,and can leverage visual information perception in the collaboration invironment.However,the fixation computation and calibration of eye tracking in the process of multiple user collaboration is more complicated,and the influence of visualization of eye movement on human visual behavior and team efficiency has not been explored yet.These challenges limit the collaborative eye tracking technique used in more applications.In this paper,an improved eye tracking technique was proposed for collaborative interaction.By analyzing the visual behavior of users in collaborative environment,a reasonable visualization form of eye movement data was designed.Finally,the relevant human-computer interaction application based on callorative eye tracking was validated with the user study.The main research work includes the following aspects:(1)Eye tracking technique based on image processing.In the case of infrared light illumination,pupil and the Purkinje image had color difference obviously.Therefore,the image processing technique was used to extract the center of the pupil and the Purkinje image.Based on the vector of the pupil center and Purkinje image center,the mapping model of pupil center cornea reflection and fixation position on the screen was created.(2)Collaborative eye tracking calibration method.As the eye movement mapping model was user dependent,it could not be applied to different users in the collaborative interaction environment.The existing eye movement mapping model was adjusted,and then the collaborative calibration mothod based on gradient optimization was proposed to compensate the individual difference,4 point calibration method was proposed and cound simplify the calibration process.It could achieve 0.69°～1.37 ° eye tracking accuracy and 52.9 frames per second for eye movement data sampling.(3)Eye trakcing data visualization for collaborative interaction.The effects of visual attributes such as color and shape on visual attention behavior were analyzed,and three specific visualization forms of eye tracking data were designed,such as fixation dots,scatters and trajectories.In view of the difficulty of understanding user's intention caused by the autonomy and ambiguity of visual behavior,the user study for visual search was carried out.Based on the evaluation with degree of cross-recurrence and fixations coverage,the results showed that fixation dots was the best one to achive highest accuracy and efficiency for the collaborative visual search tasks.(4)Human-computer interaction application based on collaborative eye tracking.Based on the client-server architecture,a collaborative eye tracking prototype system was designed and developed.The system could record and transit eye tracking data among different clients simultaneously.In view of the joint visual attention mechanism in visual attention behavior,the eye tracking data visualization was shared among multiple users in the collaborative ineracion invironment.In the case of collaborative source code peer review,fixation points,borders of code lines,gray of code line background and lines between parided code lines were shared in real time bettwen the peer reviewers during the bugs searching process.The user study results showed that,compared to no-visualization shared case,our method saved 20.1%time to find out all the bugs of the soure code peer review task.