Temporally-Consistent Surface Reconstruction using Metrically-Consistent Atlases

TL;DR

This paper introduces an unsupervised neural network-based method for reconstructing temporally consistent 3D surfaces from evolving point clouds, ensuring meaningful correspondences and robustness to noise and motion.

Contribution

It presents a novel approach using atlases and metric tensor similarity to achieve semantically meaningful, temporally consistent surface reconstructions without pre-alignment.

Findings

Outperforms state-of-the-art methods on challenging datasets

Robust to noise and global motions

Provides dense, semantically meaningful correspondences

Abstract

We propose a method for unsupervised reconstruction of a temporally-consistent sequence of surfaces from a sequence of time-evolving point clouds. It yields dense and semantically meaningful correspondences between frames. We represent the reconstructed surfaces as atlases computed by a neural network, which enables us to establish correspondences between frames. The key to making these correspondences semantically meaningful is to guarantee that the metric tensors computed at corresponding points are as similar as possible. We have devised an optimization strategy that makes our method robust to noise and global motions, without a priori correspondences or pre-alignment steps. As a result, our approach outperforms state-of-the-art ones on several challenging datasets. The code is available at https://github.com/bednarikjan/temporally_coherent_surface_reconstruction.