DualReg：面向刚性配准的双空间过滤与强化
DualReg: Dual-Space Filtering and Reinforcement for Rigid Registration
用轻量单点 RANSAC 先过滤特征对应、再以过滤后的对应为锚框提取几何代理做局部精配准，这套「特征粗配 + 几何精修」的双空间流程可直接迁移到点云配准与 SLAM 前端，兼顾大变换鲁棒性与实时性。
原文依据 we treat the filtered correspondences as anchor points, extract geometric proxies, and formulate an effective objective function
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论噪声、部分重叠的数据以及实时处理需求，给刚性配准带来了重大挑战。考虑到基于特征的匹配能够处理较大的变换差异，但精度有限；而基于局部几何的匹配能够实现细粒度的局部对齐，却严重依赖良好的初始变换，我们提出一种新的双空间范式，以充分发挥两类方法的优势。首先，我们引入一种高效的过滤机制，由计算轻量的单点 RANSAC 算法和随后的精化模块组成，用于剔除不可靠的基于特征的对应关系。随后，我们将过滤后的对应关系视为锚框点，提取几何代理，并构建一个有效的目标函数，配合专门设计的求解器来估计变换。实验验证了本方法的有效性：在 KITTI 上相比 MAC 取得 32 倍的 CPU 时间加速，且精度相当。项目主页：https://ustc3dv.github.io/DualReg/。
英文原文摘要 1052 字符
Noisy, partially overlapping data and the need for real-time processing pose major challenges for rigid registration. Considering that feature-based matching can handle large transformation differences but suffers from limited accuracy, while local geometry-based matching can achieve fine-grained local alignment but relies heavily on a good initial transformation, we propose a novel dual-space paradigm to fully leverage the strengths of both approaches. First, we introduce an efficient filtering mechanism consisting of a computationally lightweight one-point RANSAC algorithm and a subsequent refinement module to eliminate unreliable feature-based correspondences. Subsequently, we treat the filtered correspondences as anchor points, extract geometric proxies, and formulate an effective objective function with a tailored solver to estimate the transformation. Experiments verify our method's effectiveness, as demonstrated by a 32x CPU-time speedup over MAC on KITTI with comparable accuracy. Project page: https://ustc3dv.github.io/DualReg/.