发表机构
Shandong University of Science and Technology（山东科技大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对2D-3D配准在低重叠、遮挡等场景下的脆弱性，提出DMM-Align闭环框架，通过双角色扩散和可微几何铰链耦合对应细化、位姿估计与表示学习，实验验证其优于强基线。
AI 中文摘要
2D-3D配准在低重叠、遮挡、重复结构以及严重的跨模态模糊等挑战性场景中仍然脆弱。一个关键原因是现有方法孤立地改进表示学习、对应估计或位姿计算，而主要的失败模式本质上是跨层次的，错误在特征、对应关系和位姿之间传播。为解决这一局限，我们提出DMM-Align：基于扩散的匹配矩阵对齐，一种闭环框架，通过共享的可微几何状态将对应关系细化、位姿估计和表示学习耦合在一起。我们的方法利用扩散的两个协调角色：几何感知的扩散过程细化软匹配矩阵以实现鲁棒的对应估计，而几何条件扩散教师将位姿诱导的监督注入特征学习。这些过程通过一个可微的几何铰链连接，该铰链将对应关系转换为全局位姿，并将几何不一致性暴露给上游模块。在7-Scenes和RGB-D Scenes V2上的大量实验表明，DMM-Align持续优于强基线，特别是在低重叠和重度遮挡条件下，凸显了闭环几何反馈对鲁棒2D-3D配准的有效性。
英文摘要
2D-3D registration remains brittle in challenging scenarios such as low overlap, occlusion, repetitive structures, and severe cross-modal ambiguity. A key reason is that existing methods improve representation learning, correspondence estimation, or pose computation in isolation, while the dominant failure mode is inherently cross-level, where errors propagate between features, correspondences, and pose. To address this limitation, we propose DMM-Align: Diffusion-based Matching Matrix Alignment, a closed-loop framework that couples correspondence refinement, pose estimation, and representation learning through a shared differentiable geometric state. Our method leverages diffusion in two coordinated roles: a geometry-aware diffusion process refines the soft matching matrix for robust correspondence estimation, while a geometry-conditioned diffusion teacher injects pose-induced supervision back into feature learning. These processes are connected via a differentiable geometric hinge that converts correspondences into a global pose and exposes geometric inconsistency to upstream modules. Extensive experiments on 7-Scenes and RGB-D Scenes V2 demonstrate that DMM-Align consistently outperforms strong baselines, especially under low-overlap and heavy-occlusion conditions, highlighting the effectiveness of closed-loop geometric feedback for robust 2D-3D registration.
