Research map / 3D vision / Depth and stereo
A Polynomial Formula for the Perspective Four Points Problem
David Lehavi, Brian Osserman
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 |
|---|---|---|---|
| use the explicit four-point formula instead of EPnP | P4P solver · Runtime per configuration | time per configuration (us): 25.771 -> 0.477 | Improved |
| use the explicit four-point formula instead of SQPnP | P4P solver · Runtime per configuration | time per configuration (us): 36.312 -> 0.477 | Improved |
| compile with AVX2 (ablation) | P4P solver · Runtime per configuration | time per configuration (us): 0.477 -> 0.258 | Improved |
| use strict error threshold 0.05 instead of threshold 1 (ablation) | Ours · General configuration, noise 0 | rotation error (degrees): 6.7 -> 0.5 | Improved |
| use strict error threshold 0.05 instead of threshold 1 (ablation) | Ours · General configuration, noise 0 | success count: 9387 -> 7884 | Hurt |
| use ours (threshold 0.05) instead of EPnP | Ours · General configuration, noise 0 | rotation error (degrees): 11.4 -> 0.5 | Improved |
| use ours (threshold 0.05) instead of SQPnP at high noise | Ours · General configuration, noise 30 | rotation error (degrees): 5.0 -> 6.0 | Hurt |
| use ours (threshold 0.05) instead of SQPnP on collinear triples | Ours · Three collinear points, noise 0 | rotation error (degrees): 9.3 -> 2.2 | Improved |
| use ours (threshold 0.05) instead of EPnP on collinear triples | Ours · Three collinear points, noise 0 | rotation error (degrees): 43.5 -> 2.2 | Improved |
| use ours (threshold 0.05) instead of SQPnP on planar points | Ours · Planar configuration, noise 0 | rotation error (degrees): 7.2 -> 8.0 | No clear effect |
| use error-based early rejection when one 3D point is mismatched | Ours · Fast rejection experiment | rejection rate: 99% rejected at threshold .05; 96% at threshold 0.1 | Improved |
| EPnP and SQPnP on mismatched points (no rejection) | EPnP / SQPnP · Fast rejection experiment | success count: 10000 successes (EPnP), 9996 (SQPnP) | Hurt |
Closest papers by meaning
- An Efficient Algebraic Solution to the Perspective-Three-Point Problem (CVPR 2017 · similarity 0.89)
- A Consistently Fast and Globally Optimal Solution to the Perspective-n-Point Problem (ECCV 2020 · similarity 0.89)
- A Minimal Solution to the Perspective-3-Point Problem for the Camera With Unknown Focal Length and Two Degrees-of-Freedom Rotation (TPAMI 2026 · similarity 0.87)
- Efficient Solution of Point-Line Absolute Pose (CVPR 2024 · similarity 0.86)
- Three-view Focal Length Recovery From Homographies (CVPR 2025 · similarity 0.84)
- Practical Solutions to the Relative Pose of Three Calibrated Cameras (CVPR 2025 · similarity 0.84)
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.
