VAD-GS：面向动态城市场景三维高斯泼溅的可见性感知稠密化
VAD-GS: Visibility-Aware Densification for 3D Gaussian Splatting in Dynamic Urban Scenes
针对 3DGS 在无界动态城市场景中初始化点云覆盖不全导致梯度错误传播的问题，用体素级可见性推理筛出不可靠几何、再用多视图立体补出缺失结构来生成新的高斯基元，可直接迁移到自动驾驶场景重建与动态物体建模的稠密化环节。
原文依据 we propose VAD-GS, a 3DGS framework tailored for geometry recovery in challenging urban scenes
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论三维高斯泼溅（3DGS）在高保真新视角合成方面已展现出令人瞩目的性能。然而，其效果在很大程度上取决于初始化点云的质量。具体而言，要在底层场景结构上实现均匀且完整的点覆盖，需要观测视锥之间存在重叠，而这一假设在无界、动态的城市环境中往往不成立。使用部分初始化的点云训练高斯模型常常导致畸变与伪影，因为相机光线可能无法与有效表面相交，从而将错误的梯度传播到与遮挡或不可见几何相关联的高斯基元上。此外，现有的稠密化策略只是从已有基元中克隆与分裂高斯基元，无法从缺失结构中重建几何。为解决上述局限，我们提出 VAD-GS，一个面向挑战性城市场景几何恢复的 3DGS 框架。本方法通过基于体素的可见性推理识别不可靠的几何结构，通过多样性感知的视角选择挑选信息量大的支撑视角，并通过多视图立体重建恢复缺失结构。该设计使得即使在缺乏初始点的区域，也能在可靠几何先验的引导下生成新的高斯基元。在 Waymo 与 nuScenes 数据集上的大量实验表明，VAD-GS 优于最优的 3DGS 方法，并显著提升了静态与动态物体的重建几何质量。源代码将在论文发表后公开。
英文原文摘要 1651 字符
3D Gaussian splatting (3DGS) has demonstrated impressive performance in synthesizing high-fidelity novel views. Nonetheless, its effectiveness critically depends on the quality of the initialized point cloud. Specifically, achieving uniform and complete point coverage over the underlying scene structure requires overlapping observation frustums, an assumption that is often violated in unbounded, dynamic urban environments. Training Gaussian models with partially initialized point clouds often leads to distortions and artifacts, as camera rays may fail to intersect valid surfaces, resulting in incorrect gradient propagation to Gaussian primitives associated with occluded or invisible geometry. Additionally, existing densification strategies simply clone and split Gaussian primitives from existing ones, incapable of reconstructing geometry from missing structures. To address these limitations, we propose VAD-GS, a 3DGS framework tailored for geometry recovery in challenging urban scenes. Our method identifies unreliable geometry structures via voxel-based visibility reasoning, selects informative supporting views through diversity-aware view selection, and recovers missing structures via multi-view stereo reconstruction. This design enables the generation of new Gaussian primitives guided by reliable geometric priors, even in regions lacking initial points. Extensive experiments on the Waymo and nuScenes datasets demonstrate that VAD-GS outperforms state-of-the-art 3DGS approaches and significantly improves the quality of reconstructed geometry for both static and dynamic objects.Source code will be released upon publication.