3D Curvix: From Multiview 2D Edges to 3D Curve Segments


Qiwu Zhang (Brown University), Chiang-Heng Chien (Brown University), Ricardo Fabbri (Universidade do Estado do Rio de Janeiro), Benjamin Kimia (Brown University)
The 35th British Machine Vision Conference
PDFPosterVideo (Right click to download)

Abstract

The semantic reconstruction of a scene relies in part on the curvilinear structure inherent in images. The recovery of curvilinear structure is not only key to the representation of objects via ridges and other object curves but is also critical to the reconstruction from texture-poor images which lack a sufficient number of features. Prior methods advocate for the recovery of edges from multiple images which have shown very redundant in the 3D edge representation. This paper proposes 3D Curvix, a paradigm that consolidate redundant edges arising from hypotheses edge formation and multiview reconstruction. 3D Curvix organizes 3D edges by a weighted neighborhood graph which transforms parallel groups of redundant 3D edges into a one-dimensional manifold, followed by a connectivity graph construction which links 3D edges to form 3D curve segments. The result is a collection of clean, accurate 3D curves representing a sequence of dense 3D edges. Comprehensive experiments demonstrated that the proposed 3D Curvix provides remarkable 3D curve segments that can be used for a variety of real-world applications. Code is publicly available in \url{https://github.com/C-H-Chien/3D_Curvix}.

Citation

@inproceedings{Zhang_2025_BMVC,
author    = {Qiwu Zhang and Chiang-Heng Chien and Ricardo Fabbri and Benjamin Kimia},
title     = {3D Curvix: From Multiview 2D Edges to 3D Curve Segments},
booktitle = {36th British Machine Vision Conference 2025, {BMVC} 2025, Sheffield, UK, November 24-27, 2025},
publisher = {BMVA},
year      = {2025},
url       = {https://bmva-archive.org.uk/bmvc/2025/assets/papers/Paper_656/paper.pdf}
}


Copyright © 2025 The British Machine Vision Association and Society for Pattern Recognition
The British Machine Vision Conference is organised by The British Machine Vision Association and Society for Pattern Recognition. The Association is a Company limited by guarantee, No.2543446, and a non-profit-making body, registered in England and Wales as Charity No.1002307 (Registered Office: Dept. of Computer Science, Durham University, South Road, Durham, DH1 3LE, UK).

Imprint | Data Protection