The Study of Viewing Behaviors Based on Gazing Points Visualization

The interpretation of film and television features is one of the main directions of its reviews. However, the objective evaluation of the viewing behaviors still has no results. Our research is carried out on the self-developed Gazing Points Visualization Platform. We realized audiences' gazing points visualization, completed some calculations of images, compared the significant prediction with eye tracking data and drew a matching degree diagram of them. And we explored the common characteristics of content and audience interaction from the perspective of viewing behavior. We demonstrate that most of the audiences' gazing behaviors are similar and have common rules. We also find the rule that similar images can guide the similar viewing pattern. High salience corresponds to high matching. It indicates rhythm of the film. Our results and methods can provide means for audiences' cognitive evaluation and prediction, provide data and analysis basis for movie content creation. Furthermore, our platform helps to achieve closed loop collaborative evaluation.