3D reconstruction in the wild
Abstract/Contents
- Abstract
- Are existing RGB-D reconstruction pipelines ready for broad deployment? To answer this question and assist future research, I present a dataset of thorough RGB-D videos of more than 10,000 objects, produced in the wild by paid operators for the purpose of reconstruction. An evaluation of existing RGB-D reconstruction approaches reveals that they often fail on the collected data, in part due to insufficiently robust loop closure detection and global optimization. I thus present a new approach to object and indoor scene reconstruction from RGB-D video. The key idea is to combine geometric registration of fragments with robust global optimization based on line processes. Geometric registration is error-prone due to sensor noise, which leads to aliasing of geometric detail and inability to disambiguate different surfaces in the scene. The presented optimization approach disables erroneous geometric alignments even when they significantly outnumber correct ones. The objective has a least-squares form and is optimized by a high-performance pose graph solver. Experimental results demonstrate that the presented approach substantially increases the accuracy of reconstructed models on challenging real-world sequences.
Description
Type of resource | text |
---|---|
Form | electronic; electronic resource; remote |
Extent | 1 online resource. |
Publication date | 2015 |
Issuance | monographic |
Language | English |
Creators/Contributors
Associated with | Choi, Sungjoon | |
---|---|---|
Associated with | Stanford University, Department of Electrical Engineering. | |
Primary advisor | Dill, David L | |
Thesis advisor | Dill, David L | |
Thesis advisor | Koltun, Vladlen, 1980- | |
Thesis advisor | Wetzstein, Gordon | |
Advisor | Koltun, Vladlen, 1980- | |
Advisor | Wetzstein, Gordon |
Subjects
Genre | Theses |
---|
Bibliographic information
Statement of responsibility | Sungjoon Choi. |
---|---|
Note | Submitted to the Department of Electrical Engineering. |
Thesis | Thesis (Ph.D.)--Stanford University, 2015. |
Location | electronic resource |
Access conditions
- Copyright
- © 2015 by Sungjoon Choi
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...