Handbook on Leveraging Lines for Two-View Relative Pose Estimation

TL;DR

This paper introduces a hybrid approach combining points, lines, and vanishing points for robust two-view relative pose estimation, outperforming point-only methods in accuracy while maintaining efficiency.

Contribution

It presents a novel hybrid framework that integrates multiple data modalities and minimal solvers for improved pose estimation in challenging environments.

Findings

Outperforms point-based methods in accuracy (AUC@10° increased by 1-7 points)

Effective in indoor and outdoor scenarios

Maintains comparable computational speed

Abstract

We propose an approach for estimating the relative pose between calibrated image pairs by jointly exploiting points, lines, and their coincidences in a hybrid manner. We investigate all possible configurations where these data modalities can be used together and review the minimal solvers available in the literature. Our hybrid framework combines the advantages of all configurations, enabling robust and accurate estimation in challenging environments. In addition, we design a method for jointly estimating multiple vanishing point correspondences in two images, and a bundle adjustment that considers all relevant data modalities. Experiments on various indoor and outdoor datasets show that our approach outperforms point-based methods, improving AUC@10$^\circ$ by 1-7 points while running at comparable speeds. The source code of the solvers and hybrid framework will be made public.