Consistent Mesh Colors for Multi-View Reconstructed 3D Scenes

TL;DR

This paper presents a method for creating spatially consistent mesh textures from multi-view scenes without color calibration by using a color prior and re-weighting based on best-view textures, resulting in improved consistency and efficiency.

Contribution

It introduces a novel aggregation approach that leverages a color prior and re-weighting to achieve consistent mesh textures without explicit spatial optimization.

Findings

Qualitative improvements over state-of-the-art methods

More computationally efficient approach

Effective on multiple datasets including Matterport3D

Abstract

We address the issue of creating consistent mesh texture maps captured from scenes without color calibration. We find that the method for aggregation of the multiple views is crucial for creating spatially consistent meshes without the need to explicitly optimize for spatial consistency. We compute a color prior from the cross-correlation of observable view faces and the faces per view to identify an optimal per-face color. We then use this color in a re-weighting ratio for the best-view texture, which is identified by prior mesh texturing work, to create a spatial consistent texture map. Despite our method not explicitly handling spatial consistency, our results show qualitatively more consistent results than other state-of-the-art techniques while being computationally more efficient. We evaluate on prior datasets and additionally Matterport3D showing qualitative improvements.