PEGS: Physics-Event Enhanced Large Spatiotemporal Motion Reconstruction via 3D Gaussian Splatting

TL;DR

PEGS is a novel framework that combines physical priors, event stream data, and 3D Gaussian splatting to improve large-scale, fast rigid motion reconstruction with enhanced accuracy and physical consistency.

Contribution

It introduces a new integrated approach with triple-level supervision, a motion-aware training schedule, and the first RGB-Event dataset for natural fast motion reconstruction.

Findings

PEGS outperforms existing dynamic methods in motion reconstruction accuracy.

The framework effectively enforces physical plausibility and high-temporal resolution guidance.

The new dataset enables better evaluation of fast rigid motion reconstruction methods.

Abstract

Reconstruction of rigid motion over large spatiotemporal scales remains a challenging task due to limitations in modeling paradigms, severe motion blur, and insufficient physical consistency. In this work, we propose PEGS, a framework that integrates Physical priors with Event stream enhancement within a 3D Gaussian Splatting pipeline to perform deblurred target-focused modeling and motion recovery. We introduce a cohesive triple-level supervision scheme that enforces physical plausibility via an acceleration constraint, leverages event streams for high-temporal resolution guidance, and employs a Kalman regularizer to fuse multi-source observations. Furthermore, we design a motion-aware simulated annealing strategy that adaptively schedules the training process based on real-time kinematic states. We also contribute the first RGB-Event paired dataset targeting natural, fast rigid motion across diverse scenarios. Experiments show PEGS's superior performance in reconstructing motion over large spatiotemporal scales compared to mainstream dynamic methods.