Research map / Point clouds / Shape and surface completion
Learning Normals of Noisy Points by Local Gradient-Aware Surface Filtering
Qing Li, Huifang Feng, Xun Gong, Yu-Shen Liu
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 the extra loss terms (L_pd, L_sd, L_n, L_v) to the basic L_ld loss (ablation) | LGSF · FamousShape unoriented normals | RMSE: 32.96 -> 19.98 | Improved |
| add L_n and L_v losses (ablation) | LGSF · FamousShape unoriented normals | RMSE: 29.90 -> 20.00 | Improved |
| use all loss terms rather than subsets (ablation) | LGSF · FamousShape oriented normals | RMSE: full 24.07 vs best subset 24.63 | Mixed |
| add neighborhood weighted normal aggregation at inference (ablation) | LGSF · FamousShape unoriented normals | RMSE: 20.13 -> 19.98 | Improved |
| use neighborhood scale K=8 instead of 4 or 16 (ablation) | LGSF · FamousShape unoriented normals | RMSE: 21.04 / 20.71 -> 19.98 | Improved |
| unsupervised LGSF compared with supervised methods on unoriented normals | PCPNet · PCPNet unoriented normals (average) | RMSE: 12.25 vs DeepFit 11.80 and MSECNet 9.76 | Mixed |
| use LGSF instead of CAP-UDF under noise | unoriented normals · PCPNet (high noise) | RMSE: 47.64 -> 22.49 | Improved |
| use LGSF instead of CAP-UDF on noise-free data | unoriented normals · PCPNet (no noise) | RMSE: 7.59 -> 7.60 | No clear effect |
| unsupervised LGSF vs supervised denoisers | PointCleanNet denoising · 10K 3% noise | CD (x10^4): 5.984 vs PD-Flow 5.956 | Mixed |
| add richer constraints versus NeuralGF | convergence · PCPNet oriented normals | RMSE at 10,000 iterations: 19.47 -> 19.22 | Improved |
| surface reconstruction on Lounge scene | LGSF · 3D Scene Lounge | CD L2: 0.0221 (CAP-UDF) vs 0.1158 | Hurt |
Closest papers by meaning
- NeuralGF: Unsupervised Point Normal Estimation by Learning Neural Gradient Function (NeurIPS 2023 · similarity 0.93)
- HSurf-Net: Normal Estimation for 3D Point Clouds by Learning Hyper Surfaces (NeurIPS 2022 · similarity 0.91)
- Geometry and Learning Co-Supported Normal Estimation for Unstructured Point Cloud (CVPR 2020 · similarity 0.91)
- U-CAN: Unsupervised Point Cloud Denoising with Consistency-Aware Noise2Noise Matching (NeurIPS 2025 · similarity 0.89)
- 3DMambaIPF: A State Space Model for Iterative Point Cloud Filtering via Differentiable Rendering (AAAI 2025 · similarity 0.88)
- Consistent Normal Orientation for 3D Point Clouds via Least Squares on Delaunay Graph (CVPR 2025 · similarity 0.88)
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.
