Research map / 3D vision / Depth and stereo
Making Affine Correspondences Work in Camera Geometry Computation
Daniel Barath, Michal Polic, Wolfgang Förstner, Torsten Sattler, Tomas Pajdla, Zuzana Kukelova
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 cheirality sample rejection adapted to affine correspondences (ablation) | AC-based homography estimation (1AC+1PC) · HPatches | average processing time (ms): 8.6 to 7.7 ms | Improved |
| add cheirality check to point-based homography estimation (ablation) | PC-based homography (4PC) · HPatches | processing time | Improved |
| add SPRT plus uncertainty-based model rejection (ablation) | AC-based homography and fundamental matrix estimation · HPatches / TUM, KITTI, Tanks and Temples | processing time | Improved |
| add local optimization (GC-RANSAC) on inlier point locations instead of vanilla RANSAC (ablation) | AC-based fundamental/essential/homography estimation · TUM, KITTI, Tanks and Temples, HPatches | geometric error (NSGD) | Improved |
| add symmetric least squares matching refinement of affine correspondences (ablation) | affine correspondences · five image pairs, calibrated epipolar constraints | inconsistency with epipolar constraints | Improved |
| use ACs instead of PCs in vanilla RANSAC | fundamental matrix estimation · TUM, KITTI, Tanks and Temples | NSGD error | Hurt |
Closest papers by meaning
- On Making SIFT Features Affine Covariant (IJCV 2023 · similarity 0.91)
- A Complete Solution to Generalized Relative Pose Estimation From Affine Correspondences (TPAMI 2026 · similarity 0.89)
- Learning Affine Correspondences by Integrating Geometric Constraints (CVPR 2025 · similarity 0.89)
- Minimal Solvers for Relative Pose Estimation of Multi-Camera Systems using Affine Correspondences (IJCV 2022 · similarity 0.88)
- Noisy One-point Homographies are Surprisingly Good (CVPR 2024 · similarity 0.88)
- From Correspondences to Pose: Non-minimal Certifiably Optimal Relative Pose without Disambiguation (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.
