SurfaceSplat: Connecting Surface Reconstruction and Gaussian Splatting

2025-07-21

AI Summary

For surface reconstruction and novel view synthesis from sparse views, existing signed distance function (SDF)-based methods struggle to recover fine geometric details, while 3D Gaussian Splatting (3DGS) lacks global geometric consistency. This paper proposes the first bidirectional co-optimization framework integrating SDF and 3DGS: SDF models globally consistent coarse geometry, while 3DGS enables high-fidelity rendering. Geometrically guided rendering and rendering-feedback-driven detail refinement jointly refine both representations in an iterative manner. The framework thus preserves structural integrity while significantly enhancing local geometric fidelity. Extensive experiments on the DTU and MobileBrick datasets demonstrate state-of-the-art performance in both surface reconstruction accuracy (e.g., Chamfer distance, F-Score) and novel view synthesis quality (e.g., PSNR, SSIM, LPIPS), outperforming prior approaches across all metrics.

Technology Category

• Computer Vision: 3D Computer Vision
• Search and Optimization: Mixed Discrete/Continuous Search
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Surface reconstruction and novel view rendering from sparse-view images are challenging. Signed Distance Function (SDF)-based methods struggle with fine details, while 3D Gaussian Splatting (3DGS)-based approaches lack global geometry coherence. We propose a novel hybrid method that combines the strengths of both approaches: SDF captures coarse geometry to enhance 3DGS-based rendering, while newly rendered images from 3DGS refine the details of SDF for accurate surface reconstruction. As a result, our method surpasses state-of-the-art approaches in surface reconstruction and novel view synthesis on the DTU and MobileBrick datasets. Code will be released at https://github.com/Gaozihui/SurfaceSplat.

Problem

• Bridging surface reconstruction and Gaussian splatting for sparse-view images
• Overcoming SDF's detail loss and 3DGS's geometry incoherence
• Enhancing rendering and reconstruction accuracy via hybrid SDF-3DGS fusion

Innovation

• Hybrid method combining SDF and 3DGS
• SDF enhances 3DGS rendering coherence
• 3DGS refines SDF for detailed reconstruction