Are Minimal Radial Distortion Solvers Really Necessary for Relative Pose Estimation?

2025-05-01

AI Summary

In relative pose estimation, complex radial distortion minimization solvers are often assumed necessary but introduce computational overhead and implementation complexity. Method: This work proposes two lightweight alternatives: (1) a distortion-agnostic pipeline combining parameter-space sampling with undistortion followed by a standard pinhole solver; and (2) an end-to-end lightweight CNN that directly regresses distortion parameters. Additionally, we introduce DistortedPose—the first benchmark specifically designed for distorted-scene relative pose estimation—supporting cross-camera and multi-dataset (real and synthetic) evaluation. Results: Experiments show both methods match the accuracy of state-of-the-art minimal solvers while achieving significantly higher speed and simpler implementation. We empirically establish that sampling-based approaches excel in low-distortion or small-parameter-space regimes, whereas learning-based methods better handle high distortion and generalization requirements. This work provides an interpretable, deployable paradigm for distortion-aware pose estimation, challenging the necessity of intricate algebraic solvers.

Technology Category

• Intelligent Robots: State Estimation
• Computer Vision: Biometrics, Face, Gesture & Pose
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Security and Privacy: Large-scale security measurements

Abstract

Estimating the relative pose between two cameras is a fundamental step in many applications such as Structure-from-Motion. The common approach to relative pose estimation is to apply a minimal solver inside a RANSAC loop. Highly efficient solvers exist for pinhole cameras. Yet, (nearly) all cameras exhibit radial distortion. Not modeling radial distortion leads to (significantly) worse results. However, minimal radial distortion solvers are significantly more complex than pinhole solvers, both in terms of run-time and implementation efforts. This paper compares radial distortion solvers with two simple-to-implement approaches that do not use minimal radial distortion solvers: The first approach combines an efficient pinhole solver with sampled radial undistortion parameters, where the sampled parameters are used for undistortion prior to applying the pinhole solver. The second approach uses a state-of-the-art neural network to estimate the distortion parameters rather than sampling them from a set of potential values. Extensive experiments on multiple datasets, and different camera setups, show that complex minimal radial distortion solvers are not necessary in practice. We discuss under which conditions a simple sampling of radial undistortion parameters is preferable over calibrating cameras using a learning-based prior approach. Code and newly created benchmark for relative pose estimation under radial distortion are available at https://github.com/kocurvik/rdnet.

Problem

• Comparing radial distortion solvers with simpler non-minimal approaches
• Evaluating pinhole solvers with sampled undistortion parameters
• Assessing neural networks for distortion parameter estimation

Innovation

• Combines pinhole solver with sampled undistortion parameters
• Uses neural network to estimate distortion parameters
• Compares minimal solvers with simple non-minimal approaches