FUSER：前馈多视图 3D 配准 Transformer 与 SE(3)^N 扩散精化
FUSER: Feed-Forward Multiview 3D Registration Transformer and SE(3)$^N$ Diffusion Refinement
提出一种前馈多视图配准 Transformer，在统一潜空间中直接预测全局位姿，避免成对匹配；并利用 SE(3)^N 扩散精化框架对联合位姿进行去噪校正，为多视图点云配准提供了高效的新范式。
原文依据 the first feed-forward multi-view registration transformer that processes all scans jointly in a unified, compact latent space to directly predict global poses
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论多视图点云配准通常依赖大量成对匹配来构建位姿图以进行全局同步，这种计算代价高昂，且在没有整体几何约束的情况下是病态的。本文提出 FUSER，这是首个前馈多视图配准 Transformer，它在统一紧凑的潜空间中联合处理所有扫描，直接预测全局位姿，无需任何成对估计。为保持可处理性，FUSER 采用稀疏 3D CNN 将每个扫描编码为低分辨率超点特征，保留绝对平移线索，随后使用几何交替注意力模块进行高效的扫描内和扫描间推理。特别地，我们从现成的基础模型（即 π^3）中迁移 2D 注意力先验，以增强 3D 特征注意力。在 FUSER 及其估计的基础上，我们进一步引入 FUSER-DF，一个 SE(3) 扩散精化框架，通过在联合 SE(3)^N 空间上的去噪过程来校正 FUSER 的估计。这里，FUSER 作为代理多视图配准器来建模去噪器，并推导了先验条件化的 SE(3)^N 变分下界用于去噪监督。在 3DMatch 和 ScanNet 上的大量实验证实了我们的方法具有优越的配准精度和效率。
英文原文摘要 1321 字符
Registration of multiview point clouds typically depends on extensive pairwise matching to build a pose graph for global synchronization, which is computationally expensive and ill-posed without holistic geometric constraints. In this paper, we propose FUSER, the first feed-forward multi-view registration transformer that processes all scans jointly in a unified, compact latent space to directly predict global poses without any pairwise estimation. To maintain tractability, FUSER employs a sparse 3D CNN to encode each scan into low-resolution superpoint features preserving absolute translation cues, followed by a Geometric Alternating Attention module for efficient intra- and inter-scan reasoning. Particularly, we transfer 2D attention priors from off-the-shelf foundation models (i.e., \pi^3) to enhance 3D feature attention. Building upon FUSER and its estimates, we further introduce FUSER-DF, an SE(3) diffusion refinement framework to correct FUSER's estimates through a denoising process over the joint SE(3)^N space. Here, FUSER serves as a surrogate multiview register to model the denoiser, and a prior-conditioned SE(3)^N variational lower bound is derived for denoising supervision. Extensive experiments on 3DMatch and ScanNet confirm the superior registration accuracy and efficiency of our method.