Making Affine Correspondences Work in Camera Geometry Computation
Paper i proceeding, 2020

Local features e.g. SIFT and its affine and learned variants provide region-to-region rather than point-to-point correspondences. This has recently been exploited to create new minimal solvers for classical problems such as homography, essential and fundamental matrix estimation. The main advantage of such solvers is that their sample size is smaller, e.g., only two instead of four matches are required to estimate a homography. Works proposing such solvers often claim a significant improvement in run-time thanks to fewer RANSAC iterations. We show that this argument is not valid in practice if the solvers are used naively. To overcome this, we propose guidelines for effective use of region-to-region matches in the course of a full model estimation pipeline. We propose a method for refining the local feature geometries by symmetric intensity-based matching, combine uncertainty propagation inside RANSAC with preemptive model verification, show a general scheme for computing uncertainty of minimal solvers results, and adapt the sample cheirality check for homography estimation. Our experiments show that affine solvers can achieve accuracy comparable to point-based solvers at faster run-times when following our guidelines. We make code available at


Daniel Barath

Ceske Vysoke Uceni Technicke v Praze

Magyar Tudomanyos Akademia

Michal Polic

Ceske Vysoke Uceni Technicke v Praze

Wolfgang Förstner

Universität Bonn

Torsten Sattler

Chalmers, Elektroteknik, Signalbehandling och medicinsk teknik, Digitala bildsystem och bildanalys

Ceske Vysoke Uceni Technicke v Praze

Tomas Pajdla

Ceske Vysoke Uceni Technicke v Praze

Zuzana Kukelova

Ceske Vysoke Uceni Technicke v Praze

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

03029743 (ISSN) 16113349 (eISSN)

Vol. 12356 LNCS 723-740

16th European Conference on Computer Vision, ECCV 2020
Glasgow, United Kingdom,




Datorseende och robotik (autonoma system)



Mer information

Senast uppdaterat