# A Representation Separation Perspective to Correspondences-free Unsupervised 3D Point Cloud Registration

arXiv: https://arxiv.org/abs/2203.13239 · HTML: https://mlsanity.com/paper/2203.13239/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 31 references that are in the map.
- First posted on arXiv 24 Mar 2022.

## Builds on (most cited first, 19 of 19)

- [PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation](https://mlsanity.com/paper/1612.00593/) (2016)
- [PointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space](https://mlsanity.com/paper/1706.02413/) (2017)
- [Dynamic Graph CNN for Learning on Point Clouds](https://mlsanity.com/paper/1801.07829/) (2018)
- [On the Continuity of Rotation Representations in Neural Networks](https://mlsanity.com/paper/1812.07035/) (2018)
- [Deep Closest Point: Learning Representations for Point Cloud Registration](https://mlsanity.com/paper/1905.03304/) (2019)
- [PointNetLK: Robust & Efficient Point Cloud Registration using PointNet](https://mlsanity.com/paper/1903.05711/) (2019)
- [The Perfect Match: 3D Point Cloud Matching with Smoothed Densities](https://mlsanity.com/paper/1811.06879/) (2018)
- [PRNet: Self-Supervised Learning for Partial-to-Partial Registration](https://mlsanity.com/paper/1910.12240/) (2019)
- [PPF-FoldNet: Unsupervised Learning of Rotation Invariant 3D Local Descriptors](https://mlsanity.com/paper/1808.10322/) (2018)
- [Feature-metric Registration: A Fast Semi-supervised Approach for Robust Point Cloud Registration without Correspondences](https://mlsanity.com/paper/2005.01014/) (2020)
- [3DRegNet: A Deep Neural Network for 3D Point Registration](https://mlsanity.com/paper/1904.01701/) (2019)
- [DeepGMR: Learning Latent Gaussian Mixture Models for Registration](https://mlsanity.com/paper/2008.09088/) (2020)
- [Canonical Capsules: Self-Supervised Capsules in Canonical Pose](https://mlsanity.com/paper/2012.04718/) (2020)
- [3D Local Features for Direct Pairwise Registration](https://mlsanity.com/paper/1904.04281/) (2019)
- [Robust and Precise Vehicle Localization based on Multi-sensor Fusion in Diverse City Scenes](https://mlsanity.com/paper/1711.05805/) (2017)
- [DeepMapping: Unsupervised Map Estimation From Multiple Point Clouds](https://mlsanity.com/paper/1811.11397/) (2018)
- [Defining the Pose of any 3D Rigid Object and an Associated Distance](https://mlsanity.com/paper/1612.04631/) (2016)
- [GA-NET: Global Attention Network for Point Cloud Semantic Segmentation](https://mlsanity.com/paper/2107.03101/) (2021)
- [One Framework to Register Them All: PointNet Encoding for Point Cloud Alignment](https://mlsanity.com/paper/1912.05766/) (2019)

## Built on by (most cited first, 3 of 3)

- [Robust-DefReg: A Robust Deformable Point Cloud Registration Method based on Graph Convolutional Neural Networks](https://mlsanity.com/paper/2306.04701/) (2023)
- [3D Registration in 30 Years: A Survey](https://mlsanity.com/paper/2412.13735/) (2024)
- [SynBench: A Synthetic Benchmark for Non-rigid 3D Point Cloud Registration](https://mlsanity.com/paper/2409.14474/) (2024)

## Abstract (arXiv)

3D point cloud registration in remote sensing field has been greatly advanced by deep learning based methods, where the rigid transformation is either directly regressed from the two point clouds (correspondences-free approaches) or computed from the learned correspondences (correspondences-based approaches). Existing correspondences-free methods generally learn the holistic representation of the entire point cloud, which is fragile for partial and noisy point clouds. In this paper, we propose a correspondences-free unsupervised point cloud registration (UPCR) method from the representation separation perspective. First, we model the input point cloud as a combination of pose-invariant representation and pose-related representation. Second, the pose-related representation is used to learn the relative pose wrt a "latent canonical shape" for the source and target point clouds respectively. Third, the rigid transformation is obtained from the above two learned relative poses. Our method not only filters out the disturbance in pose-invariant representation but also is robust to partial-to-partial point clouds or noise. Experiments on benchmark datasets demonstrate that our unsupervised method achieves comparable if not better performance than state-of-the-art supervised registration methods.