C-GenReg: Training-Free 3D Point Cloud Registration by Multi-View-Consistent Geometry-to-Image Generation with Probabilistic Modalities Fusion

Yuval Haitman, Amit Efraim, Joseph M. Francos

CVPR 2026

C-GenReg enables training-free 3D registration by transforming point clouds into multi-view consistent RGB images, allowing Vision Foundation Model features to augment conventional 3D geometric features for more robust registration.

---

News

• Paper accepted to CVPR 2026 (Link to the paper)

---

Overview

C-GenReg is a zero-shot framework for point cloud registration that combines two complementary signals:

• a generated-RGB branch that turns geometry into multi-view consistent RGB observations and extracts dense visual correspondences with pretrained Vision Foundation Models
• a geometric branch that preserves strong registration-oriented 3D cues directly in point space
• a probabilistic Match-then-Fuse module that combines both correspondence posteriors before estimating the final rigid transformation

The method is designed to generalize across sensing domains and supports both indoor RGB-D benchmarks and real outdoor LiDAR data.

---

Repository Status

This repository is active and will be updated.

• Paper
• Project page assets
• Environment setup instructions
• Evaluation scripts

Code release is coming soon.

---

Highlights

• Training-free, plug-and-play 3D registration
• Multi-view consistent geometry-to-image generation
• Fusion of visual and geometric correspondence posteriors
• Strong zero-shot performance on 3DMatch, ScanNet, and Waymo
• Registration on real outdoor LiDAR data without relying on RGB imagery

---

Citation

If you find this work useful, please cite:

@article{haitman2026cgenreg,
  title     = {C-GenReg: Training-Free 3D Point Cloud Registration by Multi-View-Consistent Geometry-to-Image Generation with Probabilistic Modalities Fusion},
  author    = {Haitman, Yuval and Efraim, Amit and Francos, Joseph M.},
  journal   = {arXiv preprint arXiv:2604.16680},
  year      = {2026},
  doi       = {10.48550/arXiv.2604.16680},
  url       = {https://arxiv.org/abs/2604.16680}
}