High-fidelity reproduction of night townscape images based on visual characteristics

2022 Fiscal Year Final Research Report

• Project/Area Number: 17K00254
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Perceptual information processing
• Research Institution: Osaka Electro-Communication University
• Principal Investigator: Nishi Shogo 大阪電気通信大学, 情報通信工学部, 准教授 (70411478)
• Project Period (FY): 2017-04-01 – 2023-03-31
• Keywords: 低照度環境 / 視覚特性 / 画像再現 / 分光情報

Outline of Final Research Achievements

In this project, we aimed to reproduce nighttime landscape images based on visual characteristics using an image visibility model. In addition, we also developed a method for acquiring spectral information under low-light conditions, because we believe that providing high-fidelity scene images to the visibility model contributes to high-fidelity reproduction. HDR imaging is considered to be suitable for image acquisition in low-light environments. In image reproduction based on visual characteristics, we improved the defects of conventional methods and performed image reproduction for a large number of HDR images, not limited to night townscapes. The obtained images were classified according to the brightness of the scene and subjectively evaluated, and good evaluations were obtained. The night townscape images could be reproduced with high fidelity for real scenes.

Free Research Field

色彩工学

Academic Significance and Societal Importance of the Research Achievements

低照度環境下での分光情報の獲得手法を確立すること，またダイナミックレンジの観点，ならびに照度レベルによる視覚系の応答特性の変化という生理的観点からも，従来の昼光環境下での計測と再現に比べ，困難な課題に取り組む点において意義があったと考えられる。本課題への取り組みにより夜間景観の形成・整備の一助となる手法を示したが，低照度環境下での色覚シミ ュレーションへの応用が可能であるため，薄明視や暗所視環境下における安全性シミュレーションへの応用も考えられる。また見えの予測モデルの構築により，オンライン商取引やデジタルア ーカイブ，遠隔医療などの実社会への貢献も期待できる。