AI Paper Atlas

Research map / 3D vision / Depth and stereo

Minimal Solvers for Rectifying from Radially-Distorted Scales and Change of Scales

James Pritts, Zuzana Kúkelová, Viktor Larsson, Yaroslava Lochman, Ondřej Chum

IJCV 2020 · 3D vision › Depth and stereo · approach: Camera model and calibration for SfM · 10 citations (OpenAlex)

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 changedWhereMeasuredEffect
use basis sampling selection (1,000 sampled bases, keep smallest median residual) instead of standard GRevLex basis (ablation)DES and CS minimal solvers · Synthetic noiseless scenes, 1000 instanceslog10 equation residual (stability): templates 101x155 -> 133x187, 107x152 -> 154x199 (slightly larger) but significantly more stableImproved
estimate lens distortion jointly with rectification rather than assuming pinhole or fixed lambda (ablation)H222 DES solver · Synthetic scenes, distortion study lambda -5 to 0, 1px noiseRMS warp error: stable across all distortion levels; fixed-lambda solvers' error quickly increases when ground truth differsImproved
add lens distortion estimation to the solver (vs pinhole-assuming H22 DES and H22 CS) (ablation)rigid rectification solvers · Synthetic scenes, noise sweepRMS warp error: fixed-lens-distortion H22 DES and H22 CS give biased solutionsImproved
use proposed undistorting solvers instead of conjugate translation solversminimal solver proposals · Synthetic scenes, 1px noise, 5px thresholdfraction of good proposals: roughly 10% -> 30%Improved
use CS (linearized, change-of-scale) constraints instead of DES (directly encoded scale) constraints (ablation)proposed undistorting solvers · Synthetic scenes, 1px noisecumulative warp error: nearly identical performanceNo clear effect
vary number of regions per sample (222, 32 or 4 solvers) (ablation)proposed undistorting solvers · Synthetic scenes, noise sweepRMS warp errorNo clear effect
wrap solver in RANSAC over 25 minimal samplesproposed distortion-estimating solvers · Synthetic scenes, 2px noiseRMS warp error under 5px: under 5-pixel RMS warp error 75% of the timeImproved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy