ICP-3DGS：面向大规模无界场景的无SfM 3D高斯泼溅
ICP-3DGS: SfM-free 3D Gaussian Splatting for Large-scale Unbounded Scenes
AI总结：
提出ICP-3DGS方法，结合ICP与基于优化的细化实现大相机运动下的准确位姿估计，并引入基于体素的场景致密化方法，在无需SfM先验的大规模室内外场景重建与新视角合成上优于现有方法。
摘要：
近年来，NeRFs和3D Gaussian Splatting (3DGS)等神经渲染方法在场景重建和新视角合成方面取得了显著进展。然而，它们严重依赖预处理相机位姿和来自structure-from-motion (SfM)的3D结构先验，这些在户外场景中难以获取。为应对该挑战，我们提出将Iterative Closest Point (ICP)与基于优化的细化相结合，以在大相机运动下实现准确的相机位姿估计。此外，我们引入基于体素的场景致密化方法，以指导大规模场景的重建。实验表明，我们的方法ICP-3DGS在跨不同规模室内外场景的相机位姿估计和新视角合成上均优于现有方法。源代码可在https://github.com/Chenhao-Z/ICP-3DGS获取。
英文摘要：
In recent years, neural rendering methods such as NeRFs and 3D Gaussian Splatting (3DGS) have made significant progress in scene reconstruction and novel view synthesis. However, they heavily rely on preprocessed camera poses and 3D structural priors from structure-from-motion (SfM), which are challenging to obtain in outdoor scenarios. To address this challenge, we propose to incorporate Iterative Closest Point (ICP) with optimization-based refinement to achieve accurate camera pose estimation under large camera movements. Additionally, we introduce a voxel-based scene densification approach to guide the reconstruction in large-scale scenes. Experiments demonstrate that our approach ICP-3DGS outperforms existing methods in both camera pose estimation and novel view synthesis across indoor and outdoor scenes of various scales. Source code is available at https://github.com/Chenhao-Z/ICP-3DGS.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
