Structure-Aware Gaussian Splatting for Large-Scale Scene Reconstruction
作者: Weiyi Xue, Fan Lu, Chi Zhang, Tianhang Wang, Sanqing Qu, Zehan Zheng, Boyuan Zheng, Junqiao Zhao, Guang Chen
分类: cs.CV
发布日期: 2026-07-05
💡 一句话要点
提出结构感知高斯点云以解决大规模场景重建问题
🎯 匹配领域: 支柱三：空间感知与语义 (Perception & Semantics)
关键词: 3D重建 高斯点云 场景重建 计算机视觉 虚拟现实 增强现实 图像处理
📋 核心要点
- 现有方法在大规模场景重建中面临稀疏观察区域和冗余原语的问题，导致效率和质量下降。
- 本文提出SIG调度器，通过同步图像监督与高斯频率，调节训练图像分辨率和高斯密集化过程，从而解决上述问题。
- 实验结果表明，所提方法在效率和渲染质量上均显著优于现有技术，达到了最先进的性能。
📝 摘要（中文）
3D高斯点云在新视角合成中展现出显著潜力，但在大规模场景中，稀疏观察区域导致初始点过于稀疏。以低频稀疏点初始化的高斯在高频图像监督下，常引发无序的密集化和冗余原语，降低效率和质量。为此，本文提出SIG调度器，基于场景频率收敛调节训练图像分辨率和高斯密集化过程，并引入球面约束高斯以控制优化。该框架在大规模场景中实现了频率一致、几何感知和无浮动的训练，显著提升了效率和渲染质量。
🔬 方法详解
问题定义：本文旨在解决大规模场景重建中由于稀疏观察导致的高斯初始化问题，现有方法在处理低频稀疏点时常引发冗余和效率低下。
核心思路：通过将场景重建问题重新框定为信号结构恢复，提出SIG调度器，旨在根据场景频率的收敛性调节图像监督与高斯频率的同步。
技术框架：整体框架包括两个主要模块：一是计算3D表示的平均采样频率和带宽，二是基于场景频率调节训练图像分辨率和高斯密集化过程。
关键创新：引入球面约束高斯，利用初始化点云的空间先验来控制高斯优化，确保训练过程中的频率一致性和几何感知。
关键设计：在参数设置上，SIG调度器根据场景频率动态调整训练图像的分辨率，并设计了特定的损失函数以优化高斯的密集化过程。整体网络结构兼顾了效率与质量，确保了无浮动的训练效果。
🖼️ 关键图片
📊 实验亮点
实验结果显示，所提方法在大规模场景重建中，相较于基线方法在效率和渲染质量上提升了显著的幅度，具体性能数据未提供，但整体表现达到了最先进水平。
🎯 应用场景
该研究在虚拟现实、增强现实和计算机图形学等领域具有广泛的应用潜力。通过高效的场景重建技术，可以实现更真实的环境模拟和交互体验，推动相关技术的发展与应用。
📄 摘要（原文）
3D Gaussian Splatting has demonstrated remarkable potential in novel view synthesis. In contrast to small-scale scenes, large-scale scenes inevitably contain sparsely observed regions with excessively sparse initial points. In this case, supervising Gaussians initialized from low-frequency sparse points with high-frequency images often induces uncontrolled densification and redundant primitives, degrading both efficiency and quality. Intuitively, this issue can be mitigated with scheduling strategies, which can be categorized into two paradigms: modulating target signal frequency via densification and modulating sampling frequency via image resolution. However, previous scheduling strategies are primarily hardcoded, failing to perceive the convergence behavior of scene frequency. To address this, we reframe the scene reconstruction problem from the perspective of signal structure recovery and propose SIG, a novel scheduler that synchronizes image supervision with Gaussian frequencies. Specifically, we derive the average sampling frequency and bandwidth of 3D representations, and then regulate the training image resolution and the Gaussian densification process based on scene frequency convergence. Furthermore, we introduce Sphere-Constrained Gaussians, which leverage the spatial prior of initialized point clouds to control Gaussian optimization. Our framework enables frequency-consistent, geometry-aware, and floater-free training, achieving state-of-the-art performance by a substantial margin in both efficiency and rendering quality in large-scale scenes. The code is available at:this https URL