Detailed 3D Reconstruction of Whole Environment Including Shape, Illumination, and Reflectance
| Project/Area Number |
17H00744
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (A)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Perceptual information processing
|
|---|
| Research Institution | Tokyo Institute of Technology |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
岡部 孝弘 九州工業大学, 大学院情報工学研究院, 教授 (00396904)
鳥居 秋彦 東京工業大学, 工学院, 助教 (20585179)
|
|---|
| Project Period (FY) |
2017-04-01 – 2021-03-31
|
| Project Status |
Completed (Fiscal Year 2021)
|
| Budget Amount *help |
¥44,330,000 (Direct Cost: ¥34,100,000、Indirect Cost: ¥10,230,000)
Fiscal Year 2020: ¥11,180,000 (Direct Cost: ¥8,600,000、Indirect Cost: ¥2,580,000)
Fiscal Year 2019: ¥12,480,000 (Direct Cost: ¥9,600,000、Indirect Cost: ¥2,880,000)
Fiscal Year 2018: ¥12,480,000 (Direct Cost: ¥9,600,000、Indirect Cost: ¥2,880,000)
Fiscal Year 2017: ¥8,190,000 (Direct Cost: ¥6,300,000、Indirect Cost: ¥1,890,000)
|
|---|
| Keywords | マルチビューインバースレンダリング / ３次元復元 / 光源推定 / 反射率推定 / コンピュータビジョン / コンピュータービジョン |
|---|
| Outline of Final Research Achievements |
In this study, we have developed novel multi-view inverse rendering (MVIR) methods called Spectral MVIR and Polarimetric MVIR, where detailed 3D shape, illumination, and reflectance of the scene can be estimated simultaneously from multi-view input images.
Spectral MVIR enables us to estimate detailed 3D shape and precise spectral reflectance of the scene using an off-the-shelf LED bulb/projector and a standard RGB camera. Polarimetric MVIR enables us to acquire more detailed 3D shape compared with standard MVIR by exploiting the polarization information of reflected light obtained using a polarization camera.
These achievements have been published in major journals or conferences, such as The Visual Computer, ECCV, and ICCP.
|
| Academic Significance and Societal Importance of the Research Achievements |
画像を用いた３次元復元は、コンピュータビジョンの中核を担う研究課題である。本研究では、これまで独立に発展を続けてきた幾何学的３次元復元および光学的３次元復元を融合し、シーンの3D形状・光源分布・反射率を同時推定するマルチビューインバースレンダリング (MVIR)手法を確立・発展させた点に学術的な意義がある。
MVIRは画像に含まれる幾何学的および光学的特徴を最大限活かし、シーン全体の高精度な復元を実現する手法であるため、常に高度化が求められている３次元計測技術において、多岐にわたる分野での応用が期待される。
|
Report
(5 results)
Research Products
(53 results)
-
-
[Journal Article] Long-Term Visual Localization Revisited2022
Author(s)
Toft Carl、Maddern Will、Torii Akihiko、Hammarstrand Lars、Stenborg Erik、Safari Daniel、Okutomi Masatoshi、Pollefeys Marc、Sivic Josef、Pajdla Tomas、Kahl Fredrik、Sattler Torsten
-
Journal Title
IEEE Transactions on Pattern Analysis and Machine Intelligence
Volume: 44
Issue: 4
Pages: 2074-2088
DOI
Related Report
Peer Reviewed / Int'l Joint Research
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
