Objects measurements for degraded underwater scenes

• Project/Area Number: 18H03263
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Single-year Grants
• Section: 一般
• Review Section: Basic Section 61010:Perceptual information processing-related
• Research Institution: Shiga University (2021); Kyoto University
• Principal Investigator: Iiyama Masaaki 滋賀大学, データサイエンス学部, 教授 (70362415)
• Project Period (FY): 2018-04-01 – 2022-03-31
• Project Status: Completed (Fiscal Year 2021)
• Budget Amount: ¥17,290,000 (Direct Cost: ¥13,300,000、Indirect Cost: ¥3,990,000); Fiscal Year 2021: ¥2,600,000 (Direct Cost: ¥2,000,000、Indirect Cost: ¥600,000); Fiscal Year 2020: ¥3,900,000 (Direct Cost: ¥3,000,000、Indirect Cost: ¥900,000); Fiscal Year 2019: ¥4,550,000 (Direct Cost: ¥3,500,000、Indirect Cost: ¥1,050,000); Fiscal Year 2018: ¥6,240,000 (Direct Cost: ¥4,800,000、Indirect Cost: ¥1,440,000)
• Keywords: 散乱現象 / 形状計測 / 画像復元 / 深層学習 / コンピュータビジョン / ３次元計測 / 散乱 / 3次元計測

Outline of Final Research Achievements

In our project, we developed methods for measuring 3D scene in highly-degraded scene due to light scattering, especially scenes that are hard to observe objects due to heavy scattering. We focused on three major methods for shape measurement (photometric stereo, ToF, multi-view stereo) and developed methods that estimate both scattering components and scene shape/depth. We also modeled light transport model under light scattering. We developed and evaluated a Deep-learning-based method for hazy scene and confirmed the effectiveness of our method.

Academic Significance and Societal Importance of the Research Achievements

本研究は，光の散乱によって肉眼ではよく分からない程度にまで劣化した環境でも被写体の形状や被写体までの距離を計測する技術を開発した．本プロジェクトで開発した技術は将来的には水中ロボットでの利用や災害時のレスキューロボットなどに応用できる技術であり，水中での安価かつ安定した計測技術の確立に向けて重要な一歩となった．

Research Products

• [Journal Article] Simultaneous Estimation of Object Region and Depth in Participating Media Using a ToF Camera (2020)
  • Author(s): FUJIMURA Yuki、SONOGASHIRA Motoharu、IIYAMA Masaaki
  • Journal Title: IEICE Transactions on Information and Systems Volume: E103.D Issue: 3 Pages: 660-673
  • DOI: 10.1587/transinf.2019EDP7219
  • NAID: 130007804154
  • ISSN: 0916-8532, 1745-1361
  • Year and Date: 2020-03-01
  • Peer Reviewed
• [Journal Article] Photometric Stereo in Participating Media Using an Analytical Solution for Shape-Dependent Forward Scatter (2020)
  • Author(s): Fujimura Yuki、Iiyama Masaaki、Hashimoto Atsushi、Minoh Michihiko
  • Journal Title: IEEE Transactions on Pattern Analysis and Machine Intelligence Volume: 42 Issue: 3 Pages: 708-719
  • DOI: 10.1109/tpami.2018.2889088
  • Peer Reviewed
• [Presentation] Dehazing Cost Volume for Deep Multi-view Stereo in Scattering Media (2020)
  • Author(s): Yuki Fujimura, Motoharu Sonogashira, Masaaki Iiyama
  • Organizer: Asian Conference on Computer Vision
  • Int'l Joint Research
• [Presentation] 未知散乱条件下での深層学習によるMulti-view Stereo (2020)
  • Author(s): 藤村 友貴, 薗頭 元春, 飯山 将晃
  • Organizer: 情報処理学会CVIM研究会
• [Presentation] Dehazing Cost Volume for Deep Multi-view Stereo in Scattering Media (2020)
  • Author(s): Yuki Fujimura, Motoharu Sonogashira, Masaaki Iiyama
  • Organizer: 画像の認識・理解シンポジウム(MIRU)2020
• [Presentation] 散乱媒体下でのMulti-view StereoのためのDehazing Cost Volumeの提案 (2019)
  • Author(s): 藤村 友貴, 薗頭 元春, 飯山 将晃
  • Organizer: 情報処理学会コンピュータビジョンとイメージメディア研究会
• [Presentation] Defogging Kinect: Simultaneous Estimation of Object Region and Depth in Foggy Scenes (2019)
  • Author(s): Yuki Fujimura, Motoharu Sonogashira, Masaaki Iiyama
  • Organizer: 画像の認識・理解シンポジウム
• [Presentation] 近接光源下で撮影された画像からの散乱除去と深度推定 (2019)
  • Author(s): 喜島 大揮, 藤村 友貴, 薗頭 元春, 飯山 将晃
  • Organizer: 電子情報通信学会総合大会
• [Presentation] Photometric Stereo in Participating Media Considering Shape-Dependent Forward Scatter (2018)
  • Author(s): Yuki Fujimura, Masaaki Iiyama, Atsushi Hashimoto, Michihiko Minoh
  • Organizer: IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
  • Int'l Joint Research