NeuS: Learning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstruction

TL;DR

NeuS introduces a neural surface reconstruction method that leverages volume rendering and signed distance functions to achieve high-fidelity 3D reconstructions from 2D images, overcoming limitations of previous approaches.

Contribution

NeuS presents a novel volume rendering approach for neural implicit surfaces using signed distance functions, eliminating the need for mask supervision and reducing geometric bias.

Findings

NeuS outperforms state-of-the-art methods on DTU and BlendedMVS datasets.

It achieves higher quality surface reconstructions for complex objects.

NeuS effectively handles self-occlusion and thin structures.

Abstract

We present a novel neural surface reconstruction method, called NeuS, for reconstructing objects and scenes with high fidelity from 2D image inputs. Existing neural surface reconstruction approaches, such as DVR and IDR, require foreground mask as supervision, easily get trapped in local minima, and therefore struggle with the reconstruction of objects with severe self-occlusion or thin structures. Meanwhile, recent neural methods for novel view synthesis, such as NeRF and its variants, use volume rendering to produce a neural scene representation with robustness of optimization, even for highly complex objects. However, extracting high-quality surfaces from this learned implicit representation is difficult because there are not sufficient surface constraints in the representation. In NeuS, we propose to represent a surface as the zero-level set of a signed distance function (SDF) and develop a new volume rendering method to train a neural SDF representation. We observe that the conventional volume rendering method causes inherent geometric errors (i.e. bias) for surface reconstruction, and therefore propose a new formulation that is free of bias in the first order of approximation, thus leading to more accurate surface reconstruction even without the mask supervision. Experiments on the DTU dataset and the BlendedMVS dataset show that NeuS outperforms the state-of-the-arts in high-quality surface reconstruction, especially for objects and scenes with complex structures and self-occlusion.