Arbitrary point cloud upsampling via Dual Back-Projection Network

TL;DR

This paper introduces a Dual Back-Projection network for upsampling sparse and noisy 3D point clouds, effectively capturing point correlations and achieving low reconstruction errors across various upsampling factors.

Contribution

It proposes a novel Dual Back-Projection network that enhances point cloud upsampling by back projecting feature and coordinate residues, improving accuracy and generalizability.

Findings

Achieves lowest point set matching losses on benchmarks.

Effectively handles both uniform and non-uniform sparse point clouds.

Demonstrates that generative networks are not essential for non-uniform point cloud upsampling.

Abstract

Point clouds acquired from 3D sensors are usually sparse and noisy. Point cloud upsampling is an approach to increase the density of the point cloud so that detailed geometric information can be restored. In this paper, we propose a Dual Back-Projection network for point cloud upsampling (DBPnet). A Dual Back-Projection is formulated in an up-down-up manner for point cloud upsampling. It not only back projects feature residues but also coordinates residues so that the network better captures the point correlations in the feature and space domains, achieving lower reconstruction errors on both uniform and non-uniform sparse point clouds. Our proposed method is also generalizable for arbitrary upsampling tasks (e.g. 4x, 5.5x). Experimental results show that the proposed method achieves the lowest point set matching losses with respect to the benchmark. In addition, the success of our approach demonstrates that generative networks are not necessarily needed for non-uniform point clouds.