GSM-GS: Geometry-Constrained Single and Multi-view Gaussian Splatting for Surface Reconstruction
Spatial distribution comparison of Gaussian ellipsoids. This figure compares the spatial ellipsoid distributions reconstructed by the 3DGS, PGSR and Ours algorithms for the rendered scene. The 3DGS results exhibit suboptimal performance, as their Gaussian ellipsoids fail to conform closely to object surfaces. While both PGSR and ours employ thin Gaussian ellipsoids for surface approximation, our method introduces novel constraints derived from single-view and multi-view paradigms. This optimisation yields more regular ellipsoid distributions and significantly enhanced surface conformity.
Abstract
Recently, 3D Gaussian splatting has emerged as a prominent research direction owing to its ultrarapid training speed and high-fidelity rendering capabilities. However, the unstructured and irregular nature of Gaussian point clouds poses challenges to reconstruction accuracy. This limitation frequently causes high-frequency detail loss in complex surface microstructures when relying solely on routine strategies. To address this limitation, we propose GSM-GS: a synergistic optimisation framework integrating single-view adaptive sub-region weighting constraints and multi-view spatial structure refinement. For single-view optimisation, we leverage image gradient features to partition scenes into texture-rich and textureless sub-regions. The reconstruction quality is enhanced through adaptive filtering mechanisms guided by depth discrepancy features. This preserves high-weight regions while implementing a dual-branch constraint strategy tailored to regional texture variations, thereby improving geometric detail characterisation. For multi-view optimisation, we introduce a geometry-guided cross-view point cloud association method combined with a dynamic weight sampling strategy. This constructs 3D structural normal constraints across adjacent point cloud frames, effectively reinforcing multi-view consistency and reconstruction fidelity. Extensive experiments on public datasets demonstrate that our method achieves both competitive rendering quality and geometric reconstruction.
Method
GSM-GS Overview. The algorithm framework takes sparse point cloud data and image input, initialises each point cloud into a thin Gaussian ellipsoid, and processes a single-view adaptive partitioning constraint, dual-branch optimisation strategy, weight-guided dynamic sampling strategy, and cross-view geometric correlation normal constraint. The system is innovatively optimised from the perspectives of single-view and multi-view, and finally outputs high-quality reconstruction and rendering results.
Quantitative Results
Quantitative analysis of the 3D scene reconstruction accuracy of the algorithm using chamfer distance (mm) \( \downarrow \) on the DTU dataset. "Red", "Orange", and "Yellow" indicate the best, second-best, and third-best results, respectively.
Quantitative results of rendering quality for new view synthesis on the Mip-NeRF360 dataset. “Red”, “orange”, and “yellow” represented the best, second-best, and third-best results.
Quantitative Results
Quantitative comparison on DTU dataset. The diagram reveals that the same three-dimensional spatial position corresponds to different depth values under different viewing angles, indicating a deviation from theoretical expectations.
Quantitative results of rendering quality for new view synthesis on the Mip-NeRF360 dataset. “Red”, “orange”, and “yellow” represented the best, second-best, and third-best results.
More Results
DTU
BibTeX
@article{YourPaperKey2024,
title={Your Paper Title Here},
author={First Author and Second Author and Third Author},
journal={Conference/Journal Name},
year={2024},
url={https://your-domain.com/your-project-page}
}