SplitGaussian: Reconstructing Dynamic Scenes via Visual Geometry Decomposition

2025-08-06

AI Summary

In monocular dynamic scene reconstruction, entanglement between static and dynamic components causes motion leakage, geometric distortion, and temporal flickering. To address these issues, this paper proposes a decoupled spatiotemporal Gaussian representation framework. Methodologically: (1) it explicitly separates geometric and appearance modeling for static backgrounds and dynamic foregrounds; (2) it introduces an independent deformation field for the dynamic branch, enabling view- and time-dependent appearance optimization; (3) it employs a split-and-merge strategy to precisely model dynamic objects. Experiments demonstrate that our method surpasses state-of-the-art approaches in rendering quality, geometric stability, and motion separation accuracy. It significantly improves reconstruction fidelity and spatiotemporal consistency across multiple dynamic datasets, while also accelerating training convergence.

Technology Category

• Computer Vision: Motion & Tracking
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Machine Learning: Representation Learning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Reconstructing dynamic 3D scenes from monocular video remains fundamentally challenging due to the need to jointly infer motion, structure, and appearance from limited observations. Existing dynamic scene reconstruction methods based on Gaussian Splatting often entangle static and dynamic elements in a shared representation, leading to motion leakage, geometric distortions, and temporal flickering. We identify that the root cause lies in the coupled modeling of geometry and appearance across time, which hampers both stability and interpretability. To address this, we propose extbf{SplitGaussian}, a novel framework that explicitly decomposes scene representations into static and dynamic components. By decoupling motion modeling from background geometry and allowing only the dynamic branch to deform over time, our method prevents motion artifacts in static regions while supporting view- and time-dependent appearance refinement. This disentangled design not only enhances temporal consistency and reconstruction fidelity but also accelerates convergence. Extensive experiments demonstrate that SplitGaussian outperforms prior state-of-the-art methods in rendering quality, geometric stability, and motion separation.

Problem

• Reconstructing dynamic 3D scenes from monocular video
• Separating static and dynamic elements in scene representation
• Preventing motion leakage and geometric distortions in reconstructions

Innovation

• Decomposes scenes into static and dynamic components
• Decouples motion modeling from background geometry
• Enhances temporal consistency and reconstruction fidelity