Research map / Point clouds / Registration and matching
Local-To-Global Point Cloud Registration Using a Dictionary of Viewpoint Descriptors
David Avidar, David Malah, Meir Barzohar
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 |
|---|---|---|---|
| use range-image DFT viewpoint dictionary instead of dictionary clouds | Local-to-global registration · Dictionary memory per viewpoint | bytes: 2880K -> 32.4K | Improved |
| use phase correlation of range-images instead of RMSE over dictionary clouds | Dictionary search · Typical parameters | operations: 3.19 -> 0.11 (x10^6) | Improved |
| use viewpoint dictionary instead of dictionary clouds [1] | Registration accuracy · Subset of 24 local clouds | localization error and RRE | No clear effect |
| vary viewpoint grid distance dgrid within 2 m to 5 m (ablation) | Viewpoint dictionary registration · Vancouver LiDAR | localization error and RRE: dgrid in [2m, 5m] | No clear effect |
| vary angular resolution alpha within 2 to 8 degrees (ablation) | Viewpoint dictionary registration · Vancouver LiDAR | localization error and RRE: alpha in [2o, 8o] | No clear effect |
| remove GPS prior and search the whole grid, raising Ninitial 10 to 40 and Nfinal 3 to 5 (ablation) | Viewpoint dictionary registration · Vancouver LiDAR | mean registration time (s): 2sec -> 15.4sec | Hurt |
| remove GPS prior (with more candidates) (ablation) | Viewpoint dictionary registration · Vancouver LiDAR | mean localization error (m): 0.43m -> 0.44m | No clear effect |
| further downsample clouds to 2 m voxels before RMSE verification and refinement (ablation) | Viewpoint dictionary registration · Vancouver LiDAR | registration time | Improved |
| replace proposed method with FPFH feature matching | Local-to-global registration · Subset of 24 local clouds | clouds with localization error under 3 m: 24 -> at most 6 | Hurt |
Closest papers by meaning
- RegFormer: An Efficient Projection-Aware Transformer Network for Large-Scale Point Cloud Registration (ICCV 2023 · similarity 0.88)
- HRegNet: A Hierarchical Network for Large-Scale Outdoor LiDAR Point Cloud Registration (ICCV 2021 · similarity 0.87)
- HRegNet: A Hierarchical Network for Efficient and Accurate Outdoor LiDAR Point Cloud Registration (TPAMI 2023 · similarity 0.87)
- Sparse-to-Dense Matching Network for Large-Scale LiDAR Point Cloud Registration (TPAMI 2023 · similarity 0.87)
- Density-invariant Features for Distant Point Cloud Registration (ICCV 2023 · similarity 0.87)
- FreeReg: Image-to-Point Cloud Registration Leveraging Pretrained Diffusion Models and Monocular Depth Estimators (ICLR 2024 · similarity 0.86)
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.
