Research map / Learning theory and algorithms / Matrix and tensor methods
Solution Space Analysis of Essential Matrix Based on Algebraic Error Minimization
Gaku Nakano
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 |
|---|---|---|---|
| solve the new KKT conditions globally with Bertini instead of using the DLT solution | essential matrix estimation · Synthetic data, image noise 0.5-5 px | rotation/translation error | Improved |
| use KKT-GN initialized by DLT at high noise levels | KKT-GN · Synthetic data, noise sigma > 3 px | estimation error | Hurt |
| use SDP-relaxation QCQP instead of the local solvers on few correspondences | QCQP · Synthetic data, <=25 points | rotation error | Hurt |
| increase the number of points from 6 to 1000 or reduce noise | Bertini · Synthetic data | number of real solutions: 25 to 40 | No clear effect |
| replace DLT with QCQP on Entry-P10 | LO-RANSAC local optimizer · Real data seq. Entry-P10 | rotation error E_R: 0.32 -> 0.62 | Hurt |
| replace DLT with KKT-GN on Entry-P10 | LO-RANSAC local optimizer · Real data seq. Entry-P10 | translation error E_t: 1.89 -> 1.15 | Improved |
| replace DLT with KKT-GN on ModelHouse | LO-RANSAC local optimizer · Real data seq. ModelHouse | rotation error E_R: 0.81 -> 0.59 | Improved |
| replace DLT with QCQP on Corridor | LO-RANSAC local optimizer · Real data seq. Corridor | translation error E_t: 3.10 -> 2.84 | Improved |
| replace Expm-GN with KKT-GN on Corridor | LO-RANSAC local optimizer · Real data seq. Corridor | translation error E_t: 2.84 -> 2.93 | Hurt |
| replace Expm-GN with KKT-GN in terms of first-order optimality on Castle-P19 | LO-RANSAC local optimizer · Real data seq. Castle-P19 | E_KKT (log10): -16.08 -> -12.24 | Hurt |
| replace DLT with QCQP in LO-RANSAC on Castle-P19 | LO-RANSAC local optimizer · Real data seq. Castle-P19 | time (msec): 880 -> 3067 | Hurt |
| replace minEig with KKT-GN or Expm-GN in LO-RANSAC | LO-RANSAC local optimizer · Real data, number of RANSAC iterations | iterations: 1002 -> 1003 | No clear effect |
Closest papers by meaning
- Polynomial Solvers for Saturated Ideals (ICCV 2017 · similarity 0.83)
- Algebraic Variety Models for High-Rank Matrix Completion (ICML 2017 · similarity 0.82)
- Near-Optimal Weighted Matrix Completion (JMLR 2023 · similarity 0.82)
- General Low-rank Matrix Optimization: Geometric Analysis and Sharper Bounds (NeurIPS 2021 · similarity 0.82)
- Affine Rank Minimization via Asymptotic Log-Det Iteratively Reweighted Least Squares (JMLR 2025 · similarity 0.82)
- Low Rank Matrix Completion via Robust Alternating Minimization in Nearly Linear Time (ICLR 2024 · similarity 0.81)
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.
