Point cloud registration is a task to estimate the rigid transformation between two unaligned scans, which plays an important role in many computer vision applications. Previous learning-based works commonly focus on supervised registration, which have limitations in practice. Recently, with the advance of inexpensive RGB-D sensors, several learning-based works utilize RGB-D data to achieve unsupervised registration. However, most of existing unsupervised methods follow a cascaded design or fuse RGB-D data in a unidirectional manner, which do not fully exploit the complementary information in the RGB-D data. To leverage the complementary information more effectively, we propose a network implementing multi-scale bidirectional fusion between RGB images and point clouds generated from depth images. By bidirectionally fusing visual and geometric features in multi-scales, more distinctive deep features for correspondence estimation can be obtained, making our registration more accurate. Extensive experiments on ScanNet and 3DMatch demonstrate that our method achieves new state-of-the-art performance. Code will be released at https://github.com/phdymz/PointMBF

翻译：点云配准是一项估计两幅未对齐扫描之间的刚性变换的任务，在许多计算机视觉应用中发挥着重要作用。以往基于学习的工作通常专注于有监督配准，在实践中存在局限性。近年来，随着低成本RGB-D传感器的进步，一些基于学习的工作利用RGB-D数据实现了无监督配准。然而，大多数现有无监督方法采用级联设计或以单向方式融合RGB-D数据，未能充分利用RGB-D数据中的互补信息。为了更有效地利用互补信息，我们提出了一种网络，该网络在RGB图像和从深度图像生成的点云之间实现多尺度双向融合。通过在多个尺度上双向融合视觉与几何特征，可以获得更独特的深度特征用于对应估计，从而使我们的配准更加精确。在ScanNet和3DMatch上的大量实验表明，我们的方法达到了新的最先进性能。代码将发布在https://github.com/phdymz/PointMBF。