Learning general and distinctive 3D local deep descriptors for point cloud registration

TL;DR

This paper introduces a deep learning-based method for creating 3D local descriptors that are invariant to transformations and generalize well across different domains, improving point cloud registration accuracy.

Contribution

The authors propose a novel deep neural network that produces scale and rotation-invariant 3D descriptors capable of cross-domain generalization for point cloud registration.

Findings

Descriptors outperform recent methods in cross-domain tests

Achieve state-of-the-art results in same-domain benchmarks

Robust to occlusions and clutter in diverse datasets

Abstract

An effective 3D descriptor should be invariant to different geometric transformations, such as scale and rotation, robust to occlusions and clutter, and capable of generalising to different application domains. We present a simple yet effective method to learn general and distinctive 3D local descriptors that can be used to register point clouds that are captured in different domains. Point cloud patches are extracted, canonicalised with respect to their local reference frame, and encoded into scale and rotation-invariant compact descriptors by a deep neural network that is invariant to permutations of the input points. This design is what enables our descriptors to generalise across domains. We evaluate and compare our descriptors with alternative handcrafted and deep learning-based descriptors on several indoor and outdoor datasets that are reconstructed by using both RGBD sensors and laser scanners. Our descriptors outperform most recent descriptors by a large margin in terms of generalisation, and also become the state of the art in benchmarks where training and testing are performed in the same domain.