Estimating 2D Camera Motion with Hybrid Motion Basis

TL;DR

This paper introduces CamFlow, a novel framework for estimating 2D camera motion by combining geometric and stochastic motion bases, outperforming existing methods especially in complex, non-linear scenarios.

Contribution

The work proposes a hybrid motion basis approach with a probabilistic loss, improving robustness and generalization in 2D camera motion estimation.

Findings

CamFlow outperforms state-of-the-art methods in diverse scenarios.

The hybrid approach captures complex non-linear motions.

The new benchmark isolates camera motion by masking dynamic objects.

Abstract

Estimating 2D camera motion is a fundamental computer vision task that models the projection of 3D camera movements onto the 2D image plane. Current methods rely on either homography-based approaches, limited to planar scenes, or meshflow techniques that use grid-based local homographies but struggle with complex non-linear transformations. A key insight of our work is that combining flow fields from different homographies creates motion patterns that cannot be represented by any single homography. We introduce CamFlow, a novel framework that represents camera motion using hybrid motion bases: physical bases derived from camera geometry and stochastic bases for complex scenarios. Our approach includes a hybrid probabilistic loss function based on the Laplace distribution that enhances training robustness. For evaluation, we create a new benchmark by masking dynamic objects in existing optical flow datasets to isolate pure camera motion. Experiments show CamFlow outperforms state-of-the-art methods across diverse scenarios, demonstrating superior robustness and generalization in zero-shot settings. Code and datasets are available at our project page: https://lhaippp.github.io/CamFlow/.