Rocnet: 3D robust registration of points clouds using deep learning

Abstract

This paper introduces a new method for 3D point cloud registration based on deep learning. The architecture is composed of three distinct blocs: (i) an encoder with a convolutional graph-based descriptor that encodes the immediate neighbourhood of each point and an attention mechanism that encodes the variations of the surface normals. Such descriptors are refined by highlighting attention between the points of the same set (source and target) and then between the points of the two sets. (ii) a matching process that estimates a matrix of correspondences using the Sinkhorn algorithm. (iii) Finally, the rigid transformation between the two point clouds is calculated by RANSAC using the best scores of the correspondence matrix. We conduct experiments on the ModelNet40 dataset and real-world Bunny dataset, and our proposed architecture shows very promising results, outperforming state-of-the-art methods in most of the simulated configurations.