NeuRIS: Neural Reconstruction of Indoor Scenes Using Normal Priors

Abstract

Reconstructing 3D indoor scenes from 2D images is an important task in many computer vision and graphics applications. A main challenge in this task is that large texture-less areas in typical indoor scenes make existing methods struggle to produce satisfactory reconstruction results. We propose a new method, dubbed NeuRIS, for high quality reconstruction of indoor scenes. The key idea of NeuRIS is to integrate estimated normal vectors of indoor scenes as a prior in a neural rendering framework for reconstructing large texture-less shapes and, importantly, does so in an adaptive manner to also enable the reconstruction of irregular shapes with fine details. Specifically, we evaluate the faithfulness of the normal priors on-the-fly by checking the multi-view consistency of reconstruction during the optimization process. Only the normal priors accepted as faithful will be utilized for 3D reconstruction, which typically happens in regions of smooth shapes possibly with weak texture. However, for those regions with small objects or thin structures, for which the normal priors are usually unreliable, we will only rely on visual features of the input images, since such regions typically contain relatively rich visual features (e.g., shade changes and boundary contours). Extensive experiments show that NeuRIS significantly outperforms the state-of-the-art methods in terms of reconstruction quality.

Introduction

Our method is composed by two phases. In the first phase, we train a coarse model to fit both the multi-view images and the estimated normal maps by volume rendering, without any filtering strategy. In the second phase, we adaptively impose the supervision from normal priors, where two branches are performed simultaneously: in one branch we conduct a geometric quality evaluation by computing multi-view visual consistency; in the other branch, only those prior normals that pass the geometric check are accepted as proper supervisions to the rendered normals.

Comparisons on Scannet

Results on our data (by iPhone 11)

Results on Hypersim

@article{wang2022neuris,
      	title={NeuRIS: Neural Reconstruction of Indoor Scenes Using Normal Priors}, 
      	author={Wang, Jiepeng and Wang, Peng and Long, Xiaoxiao and Theobalt, Christian and Komura, Taku and Liu, Lingjie and Wang, Wenping},
	journal={arXiv preprint},
      	year={2022}
}