Recognition‐Difficulty‐Aware Hidden Images Based on Clue‐Map

Hidden images contain one or several concealed foregrounds which can be recognized with the assistance of clues preserved by artists. Experienced artists are trained for years to be skilled enough to find appropriate hidden positions for a given image. However, it is not an easy task for amateurs to quickly find these positions when they try to create satisfactory hidden images. In this paper, we present an interactive framework to suggest the hidden positions and corresponding results. The suggested results generated by our approach are sequenced according to the levels of their recognition difficulties. To this end, we propose a novel approach for assessing the levels of recognition difficulty of the hidden images and a new hidden image synthesis method that takes spatial influence into account to make the foreground harmonious with the local surroundings. During the synthesis stage, we extract the characteristics of the foreground as the clues based on the visual attention model. We validate the effectiveness of our approach by performing two user studies, including the quality of the hidden images and the suggestion accuracy.