Effects selection technique for improving visual attraction via visual saliency map

摘要

This paper proposes a selection technique for effects via a visual saliency map to improve visual attraction of an object in a target image. The proposed technique considers an optimization problem to construct an effect sequence for editing the target image by using an effect set. The set consists of effects which are frequently used by image designers and/or image editors. This optimization problem is formulated with an objective function calculated via a saliency map and a constraint condition concerning an image distance between the original image and the edited image. A heuristic method based on a greedy algorithm is applied to obtain an approximate solution. Thus, the proposed technique can improve visual attraction of the target object with suppressing a large sense of incongruity to viewers in reasonable computational time. For a validation, the proposed technique is applied to 16 examples. In addition, these results are evaluated via a subjective evaluation based on a question survey and an objective evaluation using an eye-tracking system.