Research map / 3D vision / Depth and stereo
Theory and Practice of Structure-From-Motion Using Affine Correspondences
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Estimate the essential matrix from 2 affine correspondences instead of 5 point correspondences | Essential matrix estimation for visual odometry · sequences with high outlier rates or few matches | robustness to outliers and small data sets: 2 ACs vs 5 PCs; translation RMS error never exceeds 25° for the 2-AC method, while the 5-point method frequently exceeds | Helped |
| Estimate the homography using 2 ACs rather than 4 PCs | Homography estimation with MSAC · 12-image homography estimation dataset | average MSAC iterations: 397 -> 21 | Helped |
| Use the AC-based planar segmentation metric before MSAC | 2-AC homography estimation · 12-image homography estimation dataset | average MSAC iterations: 21 -> 3 | Helped |
| Use AC compatibility constraints instead of the UFL global multi-model fitting approach | Planar segmentation · 30 image pairs from the RGB-D dataset | segmentation accuracy and computation time | Helped |
| Use 2 ACs rather than 4 PCs to generate homography hypotheses for UFL segmentation | Planar segmentation · 30 image pairs from the RGB-D dataset | segmentation accuracy | Helped |
| Use 2 ACs rather than 5 PCs for essential matrix estimation | Visual odometry on deformation-dominated sequences · two real sequences with strong deformations | number of outlier frames or failures: 6 vs 22 outliers; 4 vs 14 failures | Helped |
Closest papers by meaning
- Learning Affine Correspondences by Integrating Geometric Constraints (CVPR 2025 · similarity 0.84)
- Efficient Solution of Point-Line Absolute Pose (CVPR 2024 · similarity 0.84)
- Structure-from-Motion with a Non-Parametric Camera Model (CVPR 2025 · similarity 0.84)
- VGGSfM: Visual Geometry Grounded Deep Structure From Motion (CVPR 2024 · similarity 0.83)
- Non-Rigid Structure-from-Motion: Temporally-Smooth Procrustean Alignment and Spatially-Variant Deformation Modeling (CVPR 2024 · similarity 0.83)
- Structure-Aware Correspondence Learning for Relative Pose Estimation (CVPR 2025 · similarity 0.83)
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.