A MODEL OF HIERARCHICAL ATTENTION FOCUSING BASED ON SALIENCY MAP AND ITS APPLICATION FOR A DISPLAY METHOD FOR SMALL MOBILE DEVICES

We propose a model of hierarchical attention focusing and its application to a display method for small mobile devices. In a previous paper 1131 we pointed out that visual resolution is low when the attended area is broad and is high when the attended area is narrow. Then we proposed a model to obtain a saliency map that expresses a tendency to attract human attention [20,21]. Combining these, in this paper, we construct a model that controls attention movement, and which, in particular, predicts how attention is focused on an interest object. Based on the model, in this paper, we also propose a method to display a large-scale image on the small screens of small mobile devices, such as mobile phones. In this method, a region of interest (ROI) is detected using the saliency model, and the region is zoomed in as if attention were hierarchically focused on the region. Main novelty of this research is that there is a proposed method determines a sequence of ROIs based on a hierarchical attention focusing model. Since this sequence is expected to correspond to how human moves his or her attention successively, it is expected that when human see the generated sequence of images on a small display, he or site would easily enjoy it without sense of discomfort.