NeUDF: Leaning Neural Unsigned Distance Fields with Volume Rendering

TL;DR

NeUDF introduces a neural rendering framework using unsigned distance functions to reconstruct surfaces with arbitrary topologies from multi-view data, overcoming limitations of signed distance functions in open surfaces.

Contribution

The paper proposes a novel UDF-based neural rendering method with new weight formulations and a normal regularization strategy for open-surface reconstruction from multi-view supervision.

Findings

Outperforms state-of-the-art methods on multiple datasets.

Effectively reconstructs complex shapes with open boundaries.

Demonstrates robustness across diverse real-world objects.

Abstract

Multi-view shape reconstruction has achieved impressive progresses thanks to the latest advances in neural implicit surface rendering. However, existing methods based on signed distance function (SDF) are limited to closed surfaces, failing to reconstruct a wide range of real-world objects that contain open-surface structures. In this work, we introduce a new neural rendering framework, coded NeUDF, that can reconstruct surfaces with arbitrary topologies solely from multi-view supervision. To gain the flexibility of representing arbitrary surfaces, NeUDF leverages the unsigned distance function (UDF) as surface representation. While a naive extension of an SDF-based neural renderer cannot scale to UDF, we propose two new formulations of weight function specially tailored for UDF-based volume rendering. Furthermore, to cope with open surface rendering, where the in/out test is no longer valid, we present a dedicated normal regularization strategy to resolve the surface orientation ambiguity. We extensively evaluate our method over a number of challenging datasets, including DTU}, MGN, and Deep Fashion 3D. Experimental results demonstrate that nEudf can significantly outperform the state-of-the-art method in the task of multi-view surface reconstruction, especially for complex shapes with open boundaries.