Making Affine Correspondences Work in Camera Geometry Computation

TL;DR

This paper demonstrates how to effectively utilize affine correspondences in camera geometry estimation, improving accuracy and speed by following specific guidelines and refining local feature geometries.

Contribution

It introduces practical guidelines and methods for using affine correspondences in camera geometry, enhancing their effectiveness in real-world applications.

Findings

Affine solvers can match point-based accuracy when used properly.

Following proposed guidelines reduces RANSAC iterations and improves runtime.

The approach achieves comparable accuracy with faster computation.

Abstract

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 https://github.com/danini/affine-correspondences-for-camera-geometry.