GVGS: Gaussian Visibility-Aware Multi-View Geometry for Accurate Surface Reconstruction

2026-01-28

AI Summary

This work addresses the limited geometric accuracy of 3D Gaussian splatting in surface reconstruction, which stems from insufficient geometric supervision—multi-view constraints are often disrupted by occlusions, while monocular depth priors suffer from scale ambiguity and local inconsistencies. To overcome these issues, we propose a coarse-to-fine depth optimization framework that integrates visibility-aware multi-view geometric consistency with a progressive quadtree-guided, patch-level affine depth calibration strategy. Our approach enhances the stability of multi-view supervision through visibility-aware aggregation and mitigates scale ambiguity in monocular depth via localized affine transformations, all while preserving fine-grained geometric details. Experiments on the DTU and TNT datasets demonstrate that our method significantly outperforms existing Gaussian-based and implicit surface reconstruction approaches, achieving superior geometric fidelity.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Multi-instance/Multi-view Learning
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

3D Gaussian Splatting enables efficient optimization and high-quality rendering, yet accurate surface reconstruction remains challenging. Prior methods improve surface reconstruction by refining Gaussian depth estimates, either via multi-view geometric consistency or through monocular depth priors. However, multi-view constraints become unreliable under large geometric discrepancies, while monocular priors suffer from scale ambiguity and local inconsistency, ultimately leading to inaccurate Gaussian depth supervision. To address these limitations, we introduce a Gaussian visibility-aware multi-view geometric consistency constraint that aggregates the visibility of shared Gaussian primitives across views, enabling more accurate and stable geometric supervision. In addition, we propose a progressive quadtree-calibrated Monocular depth constraint that performs block-wise affine calibration from coarse to fine spatial scales, mitigating the scale ambiguity of depth priors while preserving fine-grained surface details. Extensive experiments on DTU and TNT datasets demonstrate consistent improvements in geometric accuracy over prior Gaussian-based and implicit surface reconstruction methods. Codes are available at an anonymous repository: https://github.com/GVGScode/GVGS.

Problem

• surface reconstruction
• 3D Gaussian Splatting
• multi-view geometry
• depth supervision
• geometric consistency

Innovation

• Gaussian Splatting
• Visibility-aware Multi-view Geometry
• Monocular Depth Calibration
• Surface Reconstruction
• Quadtree-based Affine Calibration