Research map / 3D vision / Depth and stereo
Multi-View Non-Rigid Refinement and Normal Selection for High Quality 3D Reconstruction
Sk. Mohammadul Haque, Venu Madhav Govindu
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 global multi-view non-rigid refinement plus graph-cut normal selection (OURS) versus simple normal averaging | depth-normal fusion pipeline · Synthetic sphere with ridges (12 views) | mesh-to-mesh l2 distance: 0.451 -> 0.418 | Improved |
| add global multi-view non-rigid refinement plus graph-cut normal selection (OURS) versus simple normal averaging | depth-normal fusion pipeline · Synthetic Bunny (12 views) | mesh-to-mesh l2 distance: 0.498 -> 0.497 | No clear effect |
| add global multi-view non-rigid refinement plus graph-cut normal selection (OURS) versus simple normal averaging | depth-normal fusion pipeline · Synthetic Armadillo (12 views) | mesh-to-mesh l2 distance: 0.325 -> 0.296 | Improved |
| replace priority selection (PS) with graph-cut selection and non-rigid refinement | normal selection · Synthetic Bunny (12 views) | mesh-to-mesh l2 distance: 0.531 -> 0.497 | Improved |
| replace priority selection (PS) with graph-cut selection and non-rigid refinement | normal selection · Synthetic Armadillo (12 views) | mesh-to-mesh l2 distance: 0.387 -> 0.296 | Improved |
| replace weighted-average MERGE2-3D with graph-cut selection and non-rigid refinement | normal estimation · Synthetic Bunny (12 views) | mesh-to-mesh l2 distance: 0.710 -> 0.497 | Improved |
| replace weighted-average MERGE2-3D with graph-cut selection and non-rigid refinement | normal estimation · Synthetic sphere with ridges (12 views) | mesh-to-mesh l2 distance: 0.630 -> 0.418 | Improved |
| use simple averaging instead of priority selection or MERGE2-3D weighted averaging | baseline normal fusion · Synthetic Bunny (12 views) | mesh-to-mesh l2 distance: 0.531 -> 0.498 | Improved |
| add non-rigid alignment before normal fusion (ablation) | full pipeline · Clay horse (real Kinect scans, 6 views) | visual seam-line artefacts and fine detail | Improved |
| use graph-cut selection with smoothness term instead of fixed priority ordering (ablation) | normal selection · Buddha figurine label maps (3 views) | label spatial continuity | Improved |
| MERGE2-3D weighted averaging versus plain averaging on real data | baseline normal fusion · Real Kinect scans (horse, pot) | visual detail preservation | Improved |
Closest papers by meaning
- Robust Multiview Photometric Stereo Using Planar Mesh Parameterization (TPAMI 2016 · similarity 0.88)
- Virtual Normal: Enforcing Geometric Constraints for Accurate and Robust Depth Prediction (TPAMI 2021 · similarity 0.87)
- NormalFusion: Real-Time Acquisition of Surface Normals for High-Resolution RGB-D Scanning (CVPR 2021 · similarity 0.87)
- Just Look at the Image: Viewpoint-Specific Surface Normal Prediction for Improved Multi-View Reconstruction (CVPR 2016 · similarity 0.87)
- RNb-NeuS: Reflectance and Normal-based Multi-View 3D Reconstruction (CVPR 2024 · similarity 0.87)
- SuperNormal: Neural Surface Reconstruction via Multi-View Normal Integration (CVPR 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.
