Show simple item record

dc.contributor.author
Pautrat, Rémi
dc.contributor.supervisor
Pollefeys, Marc
dc.contributor.supervisor
Larsson, Viktor
dc.contributor.supervisor
Lepetit, Vincent
dc.date.accessioned
2024-02-20T10:43:42Z
dc.date.available
2024-02-17T16:40:38Z
dc.date.available
2024-02-20T10:43:42Z
dc.date.issued
2023
dc.identifier.uri
http://hdl.handle.net/20.500.11850/660085
dc.identifier.doi
10.3929/ethz-b-000660085
dc.description.abstract
Local features are the building block of most multi-view applications such as localization and mapping, 3D reconstruction, and neural rendering, to obtain a highly accurate 3D geometry of a scene. Thus, local features are expected to be accurate, but also to be efficient and versatile to different tasks and conditions. This thesis explores some of the limitations of modern local features and offers solutions, with a particular focus on accuracy and robustness. It starts by reviewing a common issue of the widely used feature points: the trade-off between generalization and discrimination capabilities. A light-weight solution is proposed to select the most adapted invariance among several feature descriptors. Its hierarchical approach offers fine-grained robustness, while remaining compatible with both handcrafted and learned descriptors. In a second part, we tackle the challenges of the emerging line features and demonstrate their benefits in multiple tasks. We first propose a joint neural network to detect and describe line segments in images with high accuracy and robustness. Trained without ground truth labels, our line features are also equipped with a mechanism to handle partial occlusion. We then further improve the line detection with a hybrid approach combining deep learning for its robustness, and handcrafted strategies for their high accuracy. In a final part, we draw the conclusion that points and lines are complementary features and that they should be used in combination. We thus propose a joint deep matcher of points and lines, and show that leveraging the connectivity between all features is highly beneficial for a robust matching. We finally demonstrate how the combination of points and lines can be incorporated in a wide range of geometrical tasks and can boost their downstream performance.
en_US
dc.format
application/pdf
en_US
dc.language.iso
en
en_US
dc.publisher
ETH Zurich
en_US
dc.rights.uri
http://creativecommons.org/licenses/by-sa/4.0/
dc.subject
Computer Vision
en_US
dc.subject
Local Features
en_US
dc.subject
Feature Matching
en_US
dc.subject
Visual Localization
en_US
dc.title
Connecting the Dots: Combining Points and Lines for more Robust and Accurate Local Features
en_US
dc.type
Doctoral Thesis
dc.rights.license
Creative Commons Attribution-ShareAlike 4.0 International
dc.date.published
2024-02-20
ethz.size
164 p.
en_US
ethz.code.ddc
DDC - DDC::0 - Computer science, information & general works::000 - Generalities, science
en_US
ethz.identifier.diss
29746
en_US
ethz.publication.place
Zurich
en_US
ethz.publication.status
published
en_US
ethz.leitzahl
ETH Zürich::00002 - ETH Zürich::00012 - Lehre und Forschung::00007 - Departemente::02150 - Dep. Informatik / Dep. of Computer Science::02659 - Institut für Visual Computing / Institute for Visual Computing::03766 - Pollefeys, Marc / Pollefeys, Marc
en_US
ethz.date.deposited
2024-02-17T16:40:38Z
ethz.source
FORM
ethz.eth
yes
en_US
ethz.availability
Open access
en_US
ethz.rosetta.installDate
2024-02-20T10:43:45Z
ethz.rosetta.lastUpdated
2024-02-20T10:43:45Z
ethz.rosetta.versionExported
true
ethz.COinS
ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.atitle=Connecting%20the%20Dots:%20Combining%20Points%20and%20Lines%20for%20more%20Robust%20and%20Accurate%20Local%20Features&rft.date=2023&rft.au=Pautrat,%20R%C3%A9mi&rft.genre=unknown&rft.btitle=Connecting%20the%20Dots:%20Combining%20Points%20and%20Lines%20for%20more%20Robust%20and%20Accurate%20Local%20Features
 Search print copy at ETH Library

Files in this item

Thumbnail

Publication type

Show simple item record