Closed-Form Bayesian Homography Estimation Quantifies Uncertainty From Noisy Correspondences
zhenjun_zhao · x · 2026-09-16
The paper Closed-form Bayesian homography estimation from noisy point correspondences (arXiv:2609.15227):
- Problem: conventional homography estimation gives only point estimates without quantifying uncertainty, which propagates into calibration and 3D reconstruction — critical in medical imaging and autonomous driving
- Method: derives a closed-form posterior mean in homogeneous coordinates, supplemented by an iterative Bayesian approach for non-linearities; explicitly incorporates measurement uncertainty and priors, outputting a posterior over homography parameters
- Experiments: outperforms DLT under varying noise on synthetic data; image stitching confirms real-world applicability with uncertainty information
More from Research
- ArtHOI accepted to TPAMI: turns AI-generated video into supervised-free 4D articulated human-object interaction data — liuziwei7 · 2026-10-25
- Lure-TTS: 600M voice models trained for under $200, running 4x faster than real time — smirkishere · 2026-10-25
- Math arXiv paper counts diverge by field: steady growth 2022-2025, then a shift in 2026 — IgorCarron · 2026-10-25
- AI experts couple strong benchmark forecasts with low-resolution folk models of society, sparking bottlenecks debate — random_walker · 2026-10-25
- Why voice models sound natural alone but awkward in conversation: an architecture breakdown — Gauri_the_great · 2026-10-25
- OpenAI's Shyamal Anadkat recruiting founding members for computation-physics-chemistry venture — shyamalanadkat · 2026-10-25