Fast and memory-less multi modality image matching based on higher order graph matching

• Project/Area Number: 16K16087
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Perceptual information processing
• Research Institution: Tottori University
• Principal Investigator: OYAMADA Yuji 鳥取大学, 工学研究科, 助教 (30708615)
• Project Period (FY): 2016-04-01 – 2019-03-31
• Project Status: Completed (Fiscal Year 2018)
• Budget Amount: ¥4,030,000 (Direct Cost: ¥3,100,000、Indirect Cost: ¥930,000); Fiscal Year 2018: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000); Fiscal Year 2017: ¥1,690,000 (Direct Cost: ¥1,300,000、Indirect Cost: ¥390,000); Fiscal Year 2016: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
• Keywords: Image Matching / Deep Learning / 特徴量 / マッチング / コンピュータビジョン / パターン認識 / グラフマッチング / MRF最適化 / 深層学習 / 画像 / 情報システム / 幾何学 / アルゴリズム / 機械学習

Outline of Final Research Achievements

I have changed the original research plan and done the following works. Task 1 is a deep learning solution for a rigid motion estimation from 2 images. Task 2 is an application based on the previous outcome such as transparent markers and inpainting.I have done 2 presentation in international conference and received 1 Honorable Mentions for a poster presentation.
I have conducting some experiments to compare the method and state-of-the-arts on multi modality image matching.

Academic Significance and Societal Importance of the Research Achievements

画像のマッチングは複数枚の画像を比較し画像から有用な情報を抽出するために必要な前処理である．深層学習による画像のマッチングにより，既存の画像解析システム・アプリの適用先が広がると考えられる．

Research Products

• [Int'l Joint Research] Ecole Superieure d'Ingenieurs de Rennes(フランス)
• [Presentation] Texture synthesis for stable planar tracking (2019)
  • Author(s): Clement Gledel, Hideaki Uchiyama, Yuji Oyamada, Rin-ichiro Taniguchi
  • Organizer: Virtual Reality Software and Technology
  • Int'l Joint Research
• [Presentation] Transparent Random Dot Markers (2019)
  • Author(s): Hideaki Uchiyama, Yuji Oyamada
  • Organizer: International Conference on Pattern Recognition
  • Int'l Joint Research
• [Presentation] 深層畳み込みニューラルネットワークによる単一距離画像とその法線 マップを用いた物体識別 (2018)
  • Author(s): 大西 友貴, 小山田 雄仁, 三柴 数, 近藤 克哉
  • Organizer: 電子情報通信学会PRMU研究会
• [Presentation] Deep Convolutional 3D Object Classification from a Single Depth Image and Its Normal Map (2018)
  • Author(s): Yuji Oyamada, Tomotaka Ohnishi, Kazu Mishiba and Katsuya Kondo
  • Organizer: The International Workshop on Frontiers of Computer Vision
  • Int'l Joint Research
• [Remarks] Transparent Random Dot Markers
  • URL: https://sites.google.com/view/dryujioyamada/research/transparent-random-dot-markers
• [Remarks] Texture synthesis for stable planar tracking
  • URL: https://sites.google.com/view/dryujioyamada/research/texture-synthesis-for-better-trackability