Research map / Point clouds / Registration and matching
DualReg: Dual-Space Filtering and Reinforcement for Rigid Registration
Jiayi Li, Yuxin Yao, Qiuhang Lu, Juyong Zhang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add 1-point RANSAC fast filtering before refinement (ablation) | DualReg · 3DLoMatch (FPFH) | registration recall (%): 29.1 -> 41.7 | Improved |
| add 3-point weighted RANSAC refinement (ablation) | DualReg · 3DLoMatch (FPFH) | registration recall (%): 41.3 -> 41.7 | Improved |
| add dual-space optimization (ablation) | DualReg · 3DLoMatch (FCGF) | registration recall (%): 49.2 -> 60.9 | Improved |
| add geometry-based correspondences (ablation) | DualReg · 3DLoMatch (FCGF) | registration recall (%): 55.4 -> 60.9 | Improved |
| add anchor (feature-based) correspondences (ablation) | DualReg · 3DLoMatch (FCGF) | registration recall (%): 60.3 -> 60.9 | Improved |
| aggregate geometric neighborhoods as a whole instead of patch by patch (ablation) | DualReg · KITTI (FCGF) | RMSE (cm): 12.65 -> 10.78 | Improved |
| add tangential distance D_N to length consistency (ablation) | 1-point RANSAC filtering · 3DLoMatch (FPFH) | filtered inlier ratio (%): 33.00 -> 33.55 | Improved |
| add symmetry check (ablation) | 1-point RANSAC filtering · 3DLoMatch (FCGF) | filtered inlier ratio (%): 48.64 -> 52.97 | Improved |
| add 3-point refinement after fast filtering (ablation) | outlier removal · KITTI (FPFH) | filtered inlier ratio (%): 46.13 -> 84.05 | Improved |
| use probability-weighted sampling instead of classic random RANSAC in refinement (ablation) | outlier removal · KITTI (FPFH) | filtered inlier ratio (%): 84.05 -> 85.52 | Improved |
| use bijective point-to-plane tangential distance instead of angular difference of normals (ablation) | fast filtering · KITTI (FPFH) | filtered inlier ratio (%): 45.55 -> 46.13 | Improved |
| vary normal estimation neighbors K (10, 30) or use Open3D (ablation) | DualReg · 3DMatch (FCGF) | registration recall (%): 93.65 vs 93.35 / 93.16 / 93.22 | No clear effect |
Closest papers by meaning
- FilterReg: Robust and Efficient Probabilistic Point-Set Registration Using Gaussian Filter and Twist Parameterization (CVPR 2019 · similarity 0.87)
- RoReg: Pairwise Point Cloud Registration With Oriented Descriptors and Local Rotations (TPAMI 2023 · similarity 0.86)
- Coherent Point Drift Revisited for Non-Rigid Shape Matching and Registration (CVPR 2022 · similarity 0.86)
- Progressive Correspondence Regenerator for Robust 3D Registration (CVPR 2025 · similarity 0.85)
- Scalable 3D Registration via Truncated Entry-wise Absolute Residuals (CVPR 2024 · similarity 0.84)
- HybridReg: Robust 3D Point Cloud Registration with Hybrid Motions (AAAI 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
