Analysis on the validity of eye movement data during video viewing

摘要

In recent years, eye trackers have become widely applied to the collection and analyses of eye movement data in video viewing. However, some behaviors, for example, inattention during video viewing, will lead to large amounts of invalid eye movement data, which may interfere with the subsequent analyses. The main work of this paper is analyzing and processing invalid eye movement data in video viewing. Firstly, this paper introduced the concept of invalid eye movement data during video viewing process. After that, the paper analyzed the characteristics of eye movement indicators such as the spatial position of fixation points and the number of fixations. By doing that, we drew a conclusion that if users were not concentrated in video viewing, the eye movement data showed following characteristics: a lower degree of concern to significant area, fewer fixation points, longer blink duration and more frequent blinks. Finally, the application of eye movement indicators in the subjective image quality assessment was discussed. Experimental results showed that indicators discussed in the paper laid the groundwork for further identification and removal of invalid eye movement data, and provided a theoretical basis for screening the invalid eye movement data to improve the validity.