Title :
Color image modification based on visual saliency for guiding visual attention
Author :
Kokui, Tatsuhiko ; Takimoto, Hironori ; Mitsukura, Yasue ; Kishihara, Mitsuyoshi ; Okubo, Kan
Author_Institution :
Grad. Sch. of Syst. Eng., Okayama Prefectural Univ., Soja, Japan
Abstract :
In this paper, we propose a modification method of color image, for a given region in the image, to obtain the image in which the region shows the highest saliency. Generally, it is thought that the most important part of an information system that assists human activities is a natural interface with human. In addition, eye-gaze information strongly reflects the human interest or their attention. Therefore, a gaze-based interface is promising for future practical applications. In particular, if we can smoothly guide the user´s visual attention toward a target without interrupting their current visual attention, the usefulness of the gaze-based interface will be highly enhanced. In order to realize a novel interface based on visual attention, we propose a modification method for color image based on visual saliency map. The proposed method iteratively modifies the intensity and color so that the saliency inside the specified region increases while that outside the region decreases. This iteration is carried out until we obtain the image in which the specified region is most salient over the entire image. With the image obtained in this way, we can smoothly attract the user´s visual attention to the specified region, without any interruption to the user´s gaze. We evaluated the proposed method by a gaze measurement system and subjective experiments. As a result, we confirmed that the user´s visual attention is indeed attracted toward our specified region without uncomfortable feeling.
Keywords :
image colour analysis; user interfaces; color image modification; eye-gaze information; gaze measurement system; gaze-based interface; visual attention guidance; visual saliency map; Bismuth; Color; Computational modeling; Educational institutions; Image color analysis; Image quality; Visualization;
Conference_Titel :
RO-MAN, 2013 IEEE
Conference_Location :
Gyeongju
DOI :
10.1109/ROMAN.2013.6628548