Have you ever done any image editing? Many of you probably have some experience with it. For example, you might edit photos taken with friends on your smartphone to make them look nicer, or you might crop or adjust the brightness of images for a presentation in class to make them clearer. However, even if you have a strong intuitive sense of what you want to achieve, the actual process of editing can vary. At the same time, individual Kansei is very important. In other words, my sense of what is "beautiful" may not necessarily align with your sense of "beautiful."

In this research, we aim to achieve the appearance of images using only users' sensorial expressions. For these expressions, we will utilize onomatopoeia (words that phonetically imitate sounds or describe actions). In this system, for example, users can use a magical-like phrase such as, "the capybara illustaration, twinkle!" as shown in the figure below.


We are not able to use magic, so in practice, we tune the amount of image feature variation related to onomatopoeia for each user. We utilize a neighborhood search method called taboo search during this tuning process. The figure below illustrates the flow of the image editing system using onomatopoeia.


Through evaluation experiments, we found that using onomatopoeia for image editing is easier for beginners compared to manually changing the appearance of images (using software like Photoshop). However, this system is still in a prototype because the available onomatopoeia is limited and tuning requires time. Nonetheless, research continues as a first step toward image editing methods utilizing sensorial information that seem to be uncommon in the world.

Related our works
[Short Note] Hiroshi Takenouchi, Mako Chikihira, "Illustration Editing System Using Onomatopoeia," Journal of the Japanese Society for Fuzzy Theory and Intelligent Informatics, Vol.36, No.1, pp.507-511, 2024-02 (in Japanese).
[Conference Presentation] Mako Chikihira，Hiroshi Takenouchi， "Investigation of Correspondence between Onomatopoeia and Color Tones in Image Editing", Joint Symposium of the Japan Society of Kansei Engineering, Life Software, Kansei Workshop, and Jiritsu-kai 2022, A2-2, 2022-12 (Hiroshima).
*Bold: Students in this research laboratory