MVGSR: Multi-View Consistency Gaussian Splatting for Robust Surface Reconstruction

2025-03-11

AI Summary

To address floating artifacts and color distortions in 3D Gaussian Splatting (3DGS) caused by multi-view inconsistency in dynamic scenes, this paper proposes a robust surface reconstruction method. The approach employs a lightweight 3DGS architecture—without auxiliary MLP-based segmentation modules—to jointly optimize geometric accuracy and rendering fidelity. Key contributions include: (1) a novel heuristic occluder masking mechanism guided by multi-view feature consistency, enabling effective separation of dynamic objects from background clutter; (2) a Gaussian pruning and alpha-reset strategy based on per-Gaussian multi-view contribution scores, enhancing geometric stability; and (3) a multi-view consistency loss that jointly regularizes geometry and appearance. Experiments demonstrate significant suppression of rendering artifacts in complex dynamic scenes, with superior geometric precision and visual quality compared to state-of-the-art methods.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Multi-instance/Multi-view Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Security and Privacy: Data transparency and provenance

Abstract

3D Gaussian Splatting (3DGS) has gained significant attention for its high-quality rendering capabilities, ultra-fast training, and inference speeds. However, when we apply 3DGS to surface reconstruction tasks, especially in environments with dynamic objects and distractors, the method suffers from floating artifacts and color errors due to inconsistency from different viewpoints. To address this challenge, we propose Multi-View Consistency Gaussian Splatting for the domain of Robust Surface Reconstruction ( extbf{MVGSR}), which takes advantage of lightweight Gaussian models and a {heuristics-guided distractor masking} strategy for robust surface reconstruction in non-static environments. Compared to existing methods that rely on MLPs for distractor segmentation strategies, our approach separates distractors from static scene elements by comparing multi-view feature consistency, allowing us to obtain precise distractor masks early in training. Furthermore, we introduce a pruning measure based on multi-view contributions to reset transmittance, effectively reducing floating artifacts. Finally, a multi-view consistency loss is applied to achieve high-quality performance in surface reconstruction tasks. Experimental results demonstrate that MVGSR achieves competitive geometric accuracy and rendering fidelity compared to the state-of-the-art surface reconstruction algorithms. More information is available on our project page (href{https://mvgsr.github.io}{this url})

Problem

• Addresses floating artifacts in 3DGS surface reconstruction
• Improves color consistency across different viewpoints
• Enhances robustness in dynamic environments with distractors

Innovation

• Uses lightweight Gaussian models for reconstruction
• Implements heuristics-guided distractor masking strategy
• Applies multi-view consistency loss for accuracy