Impression estimation model for textures using Big Data

2021 Fiscal Year Final Research Report

• Project/Area Number: 18K11512
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 61060:Kansei informatics-related
• Research Institution: University of Nagasaki (2020-2021); Kwansei Gakuin University (2018-2019)
• Principal Investigator: Tobitani Kensuke 長崎県立大学, 情報システム学部, 准教授 (50597333)
• Co-Investigator(Kenkyū-buntansha): 片平 建史 関西学院大学, 理工学部, 講師 (40642129) ; 橋本 翔 関西学院大学, 理工学部, 助教 (80756700)
• Project Period (FY): 2018-04-01 – 2022-03-31
• Keywords: 感性的質感 / 機械学習 / 画像生成 / テクスチャ

Outline of Final Research Achievements

In this study, we proposed an image generation method for textures with desired visual sensory texture. First, (1) subjective evaluation experiments were conducted on texture images to quantify the sensory texture. Next, (2) style features were extracted using a pre-trained VGG19. Then, (3) a sensitivity evaluation model was constructed by formulating the relationship between the quantified sensory quality and the extracted style features. Finally, (4) based on the obtained model, style features were calculated when the desired emotional quality was exaggerated, and images were generated by optimization. Furthermore, the effectiveness of the method was demonstrated through validation experiments, which confirmed that the emotional quality of the generated images was significantly improved compared to the original images.

Free Research Field

感性工学

Academic Significance and Societal Importance of the Research Achievements

近年では E コマースの普及による市場環境のグローバル化に伴い，ユーザニーズや好みの多様化が進み，プロダクトのカスタマイズ化やパーソナライズ化に対する要求が高まっている．その実現に向け，人の嗜好や満足といった感性価値を的確に把握し，それらを具体的なデザインに展開する方法が注目されている．本研究により得られる成果は，直観的な素材の質感表現を可能にするという点で，人の嗜好や満足といった感性価値に基づくデザイン支援の一助になり得る．