Research map / 3D vision / Depth and stereo
Accurate Reconstruction of Oriented 3D Points using Affine Correspondences
Carolina Raposo, Joao P. Barreto
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 the proposed 2-DoF multi-view solver instead of the prior AC correction solver [7] | multi-view AC correction · Synthetic data, N=10 views, C++ | computation time: speed up of over 30× for N = 10 | Improved |
| use the proposed linear normal solver instead of [2] | multi-view normal estimation · Synthetic data, 10 views, Matlab | computation time: about 6.5× faster than [2] for 10 views | Improved |
| use the proposed linear normal solver instead of [2] | multi-view normal estimation · Synthetic data, normal angular error | angular error of normals | No clear effect |
| add AC correction before estimating the surface normal (ablation) | normal estimation · Synthetic data, sigma = 0.2, 1, 2 | angular error of normals | Improved |
| add AC correction on top of photoconsistency refinement (ablation) | normal estimation with refinement · 5 real datasets | angular error of normals | No clear effect |
| add photoconsistency refinement (Ref4DoF) to extracted ACs (ablation) | normal estimation · 5 real datasets | angular error of normals | Improved |
| use 2-DoF epipolar-constrained refinement instead of 4-DoF refinement | Ref2DoF · 5 real datasets | angular error of normals | Improved |
| use 2-DoF instead of 4-DoF refinement | refinement · Table 1, 2 to 5 views | avg time (ms): 36.3 -> 19.7 (2 views); 168.5 -> 103.6 (5 views) | Improved |
| use 3-DoF epipolar-constrained tracker instead of 6-DoF affine KLT | Track3DoF · freiburg3 nostructure notexture far | fraction of features successfully tracked: 64.8% -> 76% | Improved |
| use 3-DoF tracker instead of 6-DoF affine KLT | Track3DoF · freiburg3 nostructure notexture far | third-quartile normal error: third quartile 1.8× larger for Track6DoF; 75% of normals below 20° | Improved |
| use 3-DoF tracker instead of 6-DoF affine KLT | Track3DoF · freiburg3 nostructure notexture far | computation time: 1.25× faster, 45ms per tracklet | Improved |
Closest papers by meaning
- On Making SIFT Features Affine Covariant (IJCV 2023 · similarity 0.90)
- Theory and Practice of Structure-From-Motion Using Affine Correspondences (CVPR 2016 · similarity 0.89)
- Fast Globally Optimal Surface Normal Estimation from an Affine Correspondence (ICCV 2023 · similarity 0.87)
- Making Affine Correspondences Work in Camera Geometry Computation (ECCV 2020 · similarity 0.87)
- A Complete Solution to Generalized Relative Pose Estimation From Affine Correspondences (TPAMI 2026 · similarity 0.87)
- Learning Affine Correspondences by Integrating Geometric Constraints (CVPR 2025 · similarity 0.87)
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.
