RPM-Net: Robust Point Matching Using Learned Features

By Zi Jian Yew, Gim Hee Lee. Computer Vision and Pattern Recognition, 2020

585 citations recorded by the source corpus.

Iterative Closest Point (ICP) solves the rigid point cloud registration problem iteratively in two steps: (1) make hard assignments of spatially closest point correspondences, and then (2) find the least-squares rigid transformation. The hard assignments of closest point correspondences based on spatial distances are sensitive to the initial rigid transformation and noisy/outlier points, which often cause ICP to converge to wrong local minima. In this paper, we propose the RPM-Net -- a less sensitive to initialization and more robust deep learning-based approach for rigid point cloud registration. To this end, our network uses the differentiable Sinkhorn layer and annealing to get soft assignments of point correspondences from hybrid features learned from both spatial coordinates and local geometry. To further improve registration performance, we introduce a secondary network to predict optimal annealing parameters. Unlike some existing methods, our RPM-Net handles missing correspondences and point clouds with partial visibility. Experimental results show that our RPM-Net achieves state-of-the-art performance compared to existing non-deep learning and recent deep learning methods. Our source code is available at the project website (https://github.com/yewzijian/RPMNet).

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/06530a08301d1fb0a77f061ab1a32f06a1c6bd85

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Canonical Capsules: Unsupervised Capsules in Canonical Pose (2020, background): “…auto-encoding – against two baselines (trained in both single-category and multi-category variants): • AtlasNetV2 [13], the state-of-the-art auto-encoder which utilizes a multi-head patch-based decoder; • 3D-PointCapsNet [58], an auto-encoder for 3D point clouds that utilize a capsule architecture.”
PREDATOR: Registration of 3D Point Clouds with Low Overlap (2020, methodology): “9 shows that PREDATOR substantially increases the relative overlap and reduces the number of pairs with overlap 70% by more than 40 pp. Comparison to direct registration methods: To be able to compare PREDATOR to RPM-Net [48] and DCP [40], we resort to the synthetic, object-centric dataset they wer”
Geometric Transformer for Fast and Robust Point Cloud Registration (2022, unknown): “The first class [12,30,31,35] follows the idea of ICP [5], which iteratively establishes soft correspondences and computes the transformation with differentiable weighted SVD.”
A comprehensive survey on point cloud registration (2021, unknown): “RPMNet [111] introduces a less sensitive to initialization and more robust deep learning-based approach for rigid point cloud registration.”
CoFiNet: Reliable Coarse-to-fine Correspondences for Robust Point Cloud Registration (2021, background): “The first [13, 14, 15] follows the idea of ICP [16], where they iteratively find dense correspondences and compute pose estimation.”
SpinNet: Learning a General Surface Descriptor for 3D Point Cloud Registration (2020, background): “In contrast to traditional handcrafted descriptors, recent works [10, 68, 3, 15, 63, 61] leverage data-driven deep neural networks to learn local features from large-scale datasets.”
PointDSC: Robust Point Cloud Registration using Deep Spatial Consistency (2021, background): “Besides, end-toend registration networks [3, 62, 63, 71] have been proposed.”