STARS-GS: Structure-Aware Regularized Gaussian Splatting for Large-Scale Aerial Surface Reconstruction

📄 arXiv: 2609.03447v1 📥 PDF

作者: Bocheng Li, Wenjuan Zhang, Jie Pan. Dongxu Han, Xuesong Ma, Yiling Yao, Yaning Wang

分类: cs.CV

发布日期: 2026-09-03

💡 一句话要点

提出STARS-GS以解决大规模航空表面重建中的几何一致性问题

🎯 匹配领域: 支柱三：空间感知与语义 (Perception & Semantics) 支柱七：动作重定向 (Motion Retargeting)

关键词: 3D表面重建 高斯散射 航空摄影测量 几何一致性 结构感知 自适应正则化 城市建模

📋 核心要点

  1. 现有方法在大规模复杂场景中面临场景划分不当、几何约束不足和均匀正则化不适应等挑战。
  2. 本文提出STARS-GS框架，通过结构感知场景划分、邻域感知高斯组织和自适应表面正则化来解决这些问题。
  3. 实验结果显示，STARS-GS在表面重建中表现优异，F1-score提升约9.1%，有效提高了几何准确性和表面完整性。

📝 摘要（中文）

大规模3D表面重建是地理空间映射和城市建模的基础。尽管3D高斯散射（3DGS）在此任务中展现了潜力，但现有方法在处理复杂场景时仍面临场景划分不当、几何约束不足和均匀正则化不适应等挑战。为此，本文提出了STARS-GS框架，采用结构感知的场景划分策略、邻域感知的高斯组织和自适应表面正则化，显著提高了几何准确性和表面完整性。实验结果表明，STARS-GS在大规模航空摄影测量基准上超越了现有高斯方法，F1-score从0.640提升至0.698，提升幅度约9.1%。

🔬 方法详解

问题定义：本文旨在解决大规模航空图像的3D表面重建问题，现有方法在处理复杂场景时容易导致几何不一致和拼接伪影等问题。

核心思路：STARS-GS框架通过引入结构感知的场景划分、邻域感知的高斯组织和自适应表面正则化，旨在更好地保持场景的连续性和几何一致性。

技术框架：该框架包括三个主要模块：1) 结构感知场景划分，2) 邻域感知高斯组织，3) 自适应表面正则化，整体流程通过这些模块协同工作来优化重建效果。

关键创新：STARS-GS的创新在于其结构感知的场景划分和邻域感知的高斯组织，这些设计使得几何约束不仅关注单个高斯的属性，还考虑其邻域的组织，显著提升了重建质量。

关键设计：在参数设置上，采用自适应的正则化强度，根据局部几何特征调整，确保在结构化区域内保持几何一致性，同时在非结构化区域保留合理的变化。损失函数设计上，强调邻域高斯的几何关系，促进更好的表面拟合。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

STARS-GS在大规模航空摄影测量基准上的实验结果显示，F1-score从0.640提升至0.698，提升幅度约9.1%。这一结果表明，STARS-GS在几何准确性和表面完整性方面显著优于现有的高斯方法，展示了其在实际应用中的有效性。

🎯 应用场景

该研究在地理空间映射、城市建模和环境监测等领域具有广泛的应用潜力。通过提高大规模航空图像的表面重建质量，STARS-GS能够为城市规划、灾后评估和生态监测等提供更为精准的数据支持，推动相关领域的发展。

📄 摘要（原文）

Large-scale 3D surface reconstruction from aerial imagery is fundamental to geospatial mapping and urban modeling. Recent advances in 3D Gaussian Splatting (3DGS) have demonstrated considerable potential for this task. However, existing methods still face three major challenges in large and complex scenes: scene partitioning may split continuous scene elements across independently optimized sub-regions; geometric constraints mainly focus on the attributes of individual Gaussians while overlooking their local organization; and uniform regularization struggles to accommodate heterogeneous geometric structures. To address these issues, we propose STARS-GS, a structure-aware 3DGS framework for large-scale surface reconstruction. First, we introduce a structure-aware scene partitioning strategy that better preserves continuous scene structures during partitioning and reduces cross-region geometric inconsistencies and stitching artifacts through boundary refinement. Second, we develop neighborhood-aware Gaussian organization that extends geometric constraints from individual primitives to their neighborhood organization, encouraging Gaussians to better conform to local surface geometry. Third, we introduce adaptive surface regularization that adjusts the regularization strength according to local geometric characteristics, promoting geometric consistency in structured regions while preserving plausible variations in unstructured regions. Extensive experiments on large-scale aerial photogrammetry benchmarks demonstrate that STARS-GS consistently outperforms the evaluated Gaussian-based methods in surface reconstruction. It increases the average F1-score from 0.640 for the second-best method to 0.698, corresponding to a relative improvement of approximately 9.1\%, demonstrating effective improvements in geometric accuracy and surface completeness.