Register Any Point: Scaling 3D Point Cloud Registration by Flow Matching

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Pan, Yue, Sun, Tao, Zhu, Liyuan, Nunes, Lucas, Armeni, Iro, Behley, Jens, Stachniss, Cyrill
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866914394478215168
author Pan, Yue
Sun, Tao
Zhu, Liyuan
Nunes, Lucas
Armeni, Iro
Behley, Jens
Stachniss, Cyrill
author_facet Pan, Yue
Sun, Tao
Zhu, Liyuan
Nunes, Lucas
Armeni, Iro
Behley, Jens
Stachniss, Cyrill
contents Point cloud registration aligns multiple unposed point clouds into a common reference frame and is a core step for 3D reconstruction and robot localization without initial guess. In this work, we cast registration as conditional generation: a learned, continuous point-wise velocity field transports noisy points to a registered scene, from which the pose of each view is recovered. Unlike prior methods that perform correspondence matching to estimate pairwise transformations and then optimize a pose graph for multi-view registration, our model directly generates the registered point cloud, yielding both efficiency and point-level global consistency. By scaling the training data and conducting test-time rigidity enforcement, our approach achieves state-of-the-art results on existing pairwise registration benchmarks and on our proposed cross-domain multi-view registration benchmark. The superior zero-shot performance on this benchmark shows that our method generalizes across view counts, scene scales, and sensor modalities even with low overlap. Source code available at: https://github.com/PRBonn/RAP.
format Preprint
id arxiv_https___arxiv_org_abs_2512_01850
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Register Any Point: Scaling 3D Point Cloud Registration by Flow Matching
Pan, Yue
Sun, Tao
Zhu, Liyuan
Nunes, Lucas
Armeni, Iro
Behley, Jens
Stachniss, Cyrill
Computer Vision and Pattern Recognition
Robotics
Point cloud registration aligns multiple unposed point clouds into a common reference frame and is a core step for 3D reconstruction and robot localization without initial guess. In this work, we cast registration as conditional generation: a learned, continuous point-wise velocity field transports noisy points to a registered scene, from which the pose of each view is recovered. Unlike prior methods that perform correspondence matching to estimate pairwise transformations and then optimize a pose graph for multi-view registration, our model directly generates the registered point cloud, yielding both efficiency and point-level global consistency. By scaling the training data and conducting test-time rigidity enforcement, our approach achieves state-of-the-art results on existing pairwise registration benchmarks and on our proposed cross-domain multi-view registration benchmark. The superior zero-shot performance on this benchmark shows that our method generalizes across view counts, scene scales, and sensor modalities even with low overlap. Source code available at: https://github.com/PRBonn/RAP.
title Register Any Point: Scaling 3D Point Cloud Registration by Flow Matching
topic Computer Vision and Pattern Recognition
Robotics
url https://arxiv.org/abs/2512.01850