Neural Intrinsic Embedding for Non-rigid Point Cloud Matching

TL;DR

This paper introduces Neural Intrinsic Embedding (NIE), a novel method for non-rigid point cloud matching that captures intrinsic structure without extensive supervision or offline basis construction, improving registration accuracy.

Contribution

The paper proposes NIE, a new intrinsic embedding technique, and a weakly-supervised framework for non-rigid point cloud registration that outperforms existing methods with less supervision.

Findings

Performs on par or better than state-of-the-art methods

Does not require ground-truth correspondence labels

Eliminates need for offline basis construction

Abstract

As a primitive 3D data representation, point clouds are prevailing in 3D sensing, yet short of intrinsic structural information of the underlying objects. Such discrepancy poses great challenges on directly establishing correspondences between point clouds sampled from deformable shapes. In light of this, we propose Neural Intrinsic Embedding (NIE) to embed each vertex into a high-dimensional space in a way that respects the intrinsic structure. Based upon NIE, we further present a weakly-supervised learning framework for non-rigid point cloud registration. Unlike the prior works, we do not require expansive and sensitive off-line basis construction (e.g., eigen-decomposition of Laplacians), nor do we require ground-truth correspondence labels for supervision. We empirically show that our framework performs on par with or even better than the state-of-the-art baselines, which generally require more supervision and/or more structural geometric input.