SparseSurf: Sparse-View 3D Gaussian Splatting for Surface Reconstruction

2025-11-18

AI Summary

Under sparse-view settings, Gaussian Splatting suffers from overfitting, geometric blurring, and degraded surface reconstruction and novel-view synthesis quality. To address these issues, this paper proposes a geometry-aware Gaussian optimization framework. Its key contributions are: (1) anisotropic Gaussian primitive modeling coupled with depth regularization to enhance geometric representation fidelity; (2) a stereo geometry-texture alignment mechanism that jointly optimizes rendering quality and explicit geometric estimation; and (3) pseudo-feature-guided multi-view geometric consistency constraints, which integrate information from both training and unseen views to mitigate overfitting under sparse supervision. Evaluated on DTU, BlendedMVS, and Mip-NeRF360 benchmarks, the method achieves significant improvements in surface reconstruction accuracy and novel-view synthesis quality, establishing new state-of-the-art performance.

Technology Category

• Computer Vision: 3D Computer Vision
• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Recent advances in optimizing Gaussian Splatting for scene geometry have enabled efficient reconstruction of detailed surfaces from images. However, when input views are sparse, such optimization is prone to overfitting, leading to suboptimal reconstruction quality. Existing approaches address this challenge by employing flattened Gaussian primitives to better fit surface geometry, combined with depth regularization to alleviate geometric ambiguities under limited viewpoints. Nevertheless, the increased anisotropy inherent in flattened Gaussians exacerbates overfitting in sparse-view scenarios, hindering accurate surface fitting and degrading novel view synthesis performance. In this paper, we propose et{}, a method that reconstructs more accurate and detailed surfaces while preserving high-quality novel view rendering. Our key insight is to introduce Stereo Geometry-Texture Alignment, which bridges rendering quality and geometry estimation, thereby jointly enhancing both surface reconstruction and view synthesis. In addition, we present a Pseudo-Feature Enhanced Geometry Consistency that enforces multi-view geometric consistency by incorporating both training and unseen views, effectively mitigating overfitting caused by sparse supervision. Extensive experiments on the DTU, BlendedMVS, and Mip-NeRF360 datasets demonstrate that our method achieves the state-of-the-art performance.

Problem

• Addresses overfitting in 3D surface reconstruction from sparse views
• Improves geometric consistency using multi-view pseudo-feature enhancement
• Bridges rendering quality and geometry estimation for better reconstruction

Innovation

• Stereo Geometry-Texture Alignment bridges rendering and geometry
• Pseudo-Feature Enhanced Geometry Consistency mitigates sparse-view overfitting
• Jointly enhances surface reconstruction and novel view synthesis