Characteristics of visual saliency caused by character feature for reconstruction of saliency map model

Abstract

Visual saliency maps have been developed to estimate the bottom-up visual attention of humans. A conventional saliency map represents a bottom-up visual attention using image features such as the intensity, orientation, and color. However, it is difficult to estimate the visual attention using a conventional saliency map in the case of a top-down visual attention. In this study, we investigate the visual saliency for characters by applying still images including both characters and symbols. The experimental results indicate that characters have specific visual saliency independent of the type of language.