Research map / 3D vision / Depth and stereo
From Two Rolling Shutters to One Global Shutter
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 |
|---|---|---|---|
| Use the full 6DOF motion model instead of a restricted motion model | Two opposite-direction rolling-shutter cameras · Synthetic RS pairs | undistorted-point error | Helped |
| Fit a separate motion model per correspondence instead of one global model (ablation) | Translation solvers txy and txyz · Synthetic RS pairs | undistorted-point error | Helped |
| Use a hybrid initialization from a simpler solver, then refine with the full model (ablation) | Hybrid LO-RANSAC (v3) · Synthetic RS pairs | undistorted-point error | No clear effect |
| Assume zero baseline rather than model a nonzero baseline (ablation) | Global motion model (v2) · Synthetic RS pairs with angular velocity 15°/frame | undistortion approximation error: below 10 pixels | No clear effect |
| Use a fixed-baseline solver instead of the zero-baseline assumption (ablation) | 6DOF motion estimation · Synthetic RS pairs with increasing baseline | solver performance stability | Helped |
| Use opposite rolling-shutter directions on two nearby cameras | Rolling-shutter image correction · Real scenes with varied motion | visual undistortion quality | Helped |
| Use optical-flow correspondences for motion estimation rather than relying only on sparse feature matches | 6DOF motion estimation for translational footage · Real translational-motion scenes | motion-estimation reliability | Helped |
| Preselect a balanced set of optical-flow correspondences instead of sampling uniformly across the image | RANSAC motion estimation · Real translational-motion scenes | motion-estimation accuracy | Helped |
| Use the txy solver for pure, fast translation instead of a general solver | Two-camera rolling-shutter correction · Pure translation orthogonal to the viewing direction | correction robustness | Helped |
| Use a global translation model to correct rotational distortions | txy and txyz solvers · Synthetic RS pairs with rotational distortion | undistorted-point error | Hurt |
| Use a local txy model to approximate rotational distortion piece by piece | txy solver · Synthetic RS pairs | undistorted-point error variance | Helped |
| Use learned single-image undistortion methods instead of the two-camera setup | Rolling-shutter image correction · Real scenes with varied motion | visual undistortion quality | Hurt |
Closest papers by meaning
- Learning Structure-And-Motion-Aware Rolling Shutter Correction (CVPR 2019 · similarity 0.90)
- Rolling Shutter Correction with Intermediate Distortion Flow Estimation (CVPR 2024 · similarity 0.89)
- Order-One Rolling Shutter Cameras (CVPR 2025 · similarity 0.88)
- Single Image Rolling Shutter Removal with Diffusion Models (AAAI 2025 · similarity 0.87)
- USB-NeRF: Unrolling Shutter Bundle Adjusted Neural Radiance Fields (ICLR 2024 · similarity 0.86)
- Motion Blur Decomposition with Cross-shutter Guidance (CVPR 2024 · similarity 0.85)
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.