Annotation has played an important role to communicate user's experience or knowledge to another user. Adding such annotation to visualized computational results in CAE has not been adopted owing to convincing users that images visualized by a user represent some knowledge itself. However, as complexity of computational data increases, interpretation of visualized results becomes more difficult, and expertise annotation is of nessecity requiered. In this paper, we propose an effective annotation method for images from CAE to support the easy interpretation of the results First, we examine user's viewpoint on a certain image to be interpreted, through eye movement analysis or mouse movement analysis. Second, characteristic features of viewpoint movement are extracted, and pattern of viewpoint dependence is inquired using a picture of the "attention map" and a chart of time variation in movement of attention point. These are added to the result as annotation.