NeRSP: Neural 3D Reconstruction for Reflective Objects with Sparse Polarized Images

TL;DR

NeRSP introduces a neural 3D reconstruction method that effectively handles reflective surfaces using sparse polarized images, leveraging polarization cues and multiview consistency to produce accurate 3D models with minimal input views.

Contribution

The paper presents a novel approach combining polarized image cues and neural implicit representations to improve 3D reconstruction of reflective objects from sparse views.

Findings

Achieves state-of-the-art results with only 6 views.

Effectively reconstructs reflective surfaces despite sparse and view-dependent challenges.

Leverages polarization and azimuth consistency for improved accuracy.

Abstract

We present NeRSP, a Neural 3D reconstruction technique for Reflective surfaces with Sparse Polarized images. Reflective surface reconstruction is extremely challenging as specular reflections are view-dependent and thus violate the multiview consistency for multiview stereo. On the other hand, sparse image inputs, as a practical capture setting, commonly cause incomplete or distorted results due to the lack of correspondence matching. This paper jointly handles the challenges from sparse inputs and reflective surfaces by leveraging polarized images. We derive photometric and geometric cues from the polarimetric image formation model and multiview azimuth consistency, which jointly optimize the surface geometry modeled via implicit neural representation. Based on the experiments on our synthetic and real datasets, we achieve the state-of-the-art surface reconstruction results with only 6 views as input.