用于快速且鲁棒点云配准的 Geometric Transformer
Geometric Transformer for Fast and Robust Point Cloud Registration
AI总结：
本文提出 Geometric Transformer，通过编码成对距离与三元组角度学习几何结构特征，实现低重叠场景下鲁棒超点匹配，并无需 RANSAC，显著提升配准速度与精度。
摘要：
我们研究为点云配准提取准确对应关系的问题。近期的无关键点方法绕过了在低重叠场景中难以实现的可重复关键点检测，在配准中展现出巨大潜力。它们在下采样的超点上寻找对应关系，然后将对应关系传播到稠密点。超点根据其相邻块是否重叠进行匹配。这种稀疏且松散的匹配需要能够捕捉点云几何结构的上下文特征。我们提出 Geometric Transformer 来学习几何特征，以实现鲁棒的超点匹配。它编码成对距离和三元组角度，使其在低重叠情况下具有鲁棒性，并且对刚性变换保持不变。这一简洁设计取得了惊人的高匹配精度，因此在估计对齐变换时不需要 RANSAC，从而带来 100 倍加速。在具有挑战性的 3DLoMatch 基准上，我们的方法将内点率提高了 17～30 个百分点，并将配准召回率提高了 7 个点以上。我们的代码和模型可在 https://github.com/qinzheng93/GeoTransformer 获取。
英文摘要：
We study the problem of extracting accurate correspondences for point cloud registration. Recent keypoint-free methods bypass the detection of repeatable keypoints which is difficult in low-overlap scenarios, showing great potential in registration. They seek correspondences over downsampled superpoints, which are then propagated to dense points. Superpoints are matched based on whether their neighboring patches overlap. Such sparse and loose matching requires contextual features capturing the geometric structure of the point clouds. We propose Geometric Transformer to learn geometric feature for robust superpoint matching. It encodes pair-wise distances and triplet-wise angles, making it robust in low-overlap cases and invariant to rigid transformation. The simplistic design attains surprisingly high matching accuracy such that no RANSAC is required in the estimation of alignment transformation, leading to $100$ times acceleration. Our method improves the inlier ratio by $17{\sim}30$ percentage points and the registration recall by over $7$ points on the challenging 3DLoMatch benchmark. Our code and models are available at https://github.com/qinzheng93/GeoTransformer.
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。
