VoxNeuS: Enhancing Voxel-Based Neural Surface Reconstruction via Gradient Interpolation

TL;DR

VoxNeuS introduces a gradient interpolation technique to improve voxel-based neural surface reconstruction, achieving higher quality results efficiently and disentangling geometry from radiance to reduce artifacts.

Contribution

The paper proposes gradient interpolation to address discontinuity issues in voxel-based SDF reconstruction and introduces a geometry-radiance disentangled architecture for better accuracy.

Findings

Achieves superior reconstruction quality over previous methods.

Training completes in 15 minutes using less than 3 GB memory.

Effectively disentangles geometry and radiance, reducing artifacts.

Abstract

Neural Surface Reconstruction learns a Signed Distance Field~(SDF) to reconstruct the 3D model from multi-view images. Previous works adopt voxel-based explicit representation to improve efficiency. However, they ignored the gradient instability of interpolation in the voxel grid, leading to degradation on convergence and smoothness. Besides, previous works entangled the optimization of geometry and radiance, which leads to the deformation of geometry to explain radiance, causing artifacts when reconstructing textured planes. In this work, we reveal that the instability of gradient comes from its discontinuity during trilinear interpolation, and propose to use the interpolated gradient instead of the original analytical gradient to eliminate the discontinuity. Based on gradient interpolation, we propose VoxNeuS, a lightweight surface reconstruction method for computational and memory efficient neural surface reconstruction. Thanks to the explicit representation, the gradient of regularization terms, i.e. Eikonal and curvature loss, are directly solved, avoiding computation and memory-access overhead. Further, VoxNeuS adopts a geometry-radiance disentangled architecture to handle the geometry deformation from radiance optimization. The experimental results show that VoxNeuS achieves better reconstruction quality than previous works. The entire training process takes 15 minutes and less than 3 GB of memory on a single 2080ti GPU.