DecomVoxel：利用3D原生先验与引导式原位去噪优化实现分解式场景重建
DecomVoxel: Harnessing 3D-Native Priors with Guided In-situ Denoising Optimization for Decompositional Scene Reconstruction
浏览论文内容
中文总结 AI 辅助
提出DecomVoxel，将物体补全表述为引导式原位去噪优化，连接3D原生先验与神经场景重建，通过epsilon蒸馏损失和自适应空间引导抑制幻觉与漂移，在Replica和ScanNet++上显著优于现有方法。
中文摘要 AI 辅助
分解式场景重建旨在重建高质量的物体和背景，然而现有方法在重度遮挡下仍难以达到理想的质量水平。虽然生成式先验提供了一种潜在解决方案，但基于2D图像的先验由于缺乏3D感知，常常遭受多视角不一致的问题。相反，3D原生先验提供了更强的结构归纳偏置，但在复杂场景中经常导致空间漂移和错位。为解决这些问题，我们提出了DecomVoxel，将物体补全表述为一种引导式原位去噪优化，从而将3D原生先验与神经场景重建相连接。我们的框架引入了一种重新表述的基于epsilon的蒸馏损失，以确保稳定的潜在空间细化，同时采用自适应空间引导，利用占据和空置锚点并结合时间退火来抑制生成式幻觉并缓解空间漂移。在Replica和ScanNet++上的实验表明，DecomVoxel显著优于最先进的方法，同时忠实地保留了原始空间布局、结构保真度和风格一致的纹理。我们的方法通过提供具有干净拓扑、几何和外观的高质量纹理网格，推动了分解式重建的边界，为复杂真实世界场景的分解式重建提供了稳健的解决方案。代码可在以下网址获取：https://this https URL。
英文摘要
Decompositional scene reconstruction aims to reconstruct high-quality objects and background, yet existing methods still struggle with the level of quality under heavy occlusions. While generative priors offer a potential solution, 2D image-based priors often suffer from multi-view inconsistency due to a lack of 3D awareness. Conversely, 3D-native priors provide stronger structural inductive biases but frequently lead to spatial drift and misalignment within complex scenes. To address these issues, we propose DecomVoxel, formulating object completion as a guided in-situ denoising optimization that bridges 3D-native priors with neural scene reconstruction. Our framework introduces a reformulated epsilon-based distillation loss to ensure stable latent refinement, alongside adaptive spatial guidance that utilizes occupied and vacant anchors with temporal annealing to suppress generative hallucinations and mitigate spatial drift. Experiments on Replica and ScanNet++ show that DecomVoxel significantly outperforms state-of-the-art methods while faithfully preserving the original spatial layout, structural fidelity, and style-consistent texture. Our method pushes the boundary of decompositional reconstruction by delivering high-quality textured meshes with clean topology, geometry, and appearance, providing a robust solution for the decompositional reconstruction of complex real-world scenes. Code is available at https://github.com/DecomVoxel/DecomVoxel.
发表机构
- Tsinghua University（清华大学）
- State Key Laboratory of General Artificial Intelligence（通用人工智能全国重点实验室）
- BIGAI（北京通用人工智能研究院）
- Peking University（北京大学）
机构由 AI 辅助整理，请以论文原文为准。
