Enhancing infographics based on symmetry saliency

Image saliency is a biologically inspired concept for characterizing visual conspicuity of individual features in natural images, and provides us with a useful insight into the mechanism for directing instant visual attention from viewers. Nevertheless, this perceptual quality often remains to be further sophisticated especially for enhancing saliency in infographic images since they usually consist of relatively simple visual pattens that result in sharp image edges rather than smooth gradations in natural images. This paper presents a new approach to intentionally drawing visual attention for infographic images, in such a way that the corresponding important features naturally pop up in the image. The idea behind our approach is to introduce the concept of symmetry saliency for enhancing local symmetry inherent in such infographic images. This is accomplished by evaluating how much each image edge contributes to the symmetry saliency, and augmenting the corresponding image gradient in proportion to the amount of its contribution. The intensity field of the given image is then modulated with such enhanced image edges by solving the Poisson equation. Several examples together with statistics obtained through a user study demonstrate that our proposed approach successfully improves the readability of infographic images and effectively attracts visual attention to intended regions of interest.