Improving Multi-View Stereo via Super-Resolution

TL;DR

This paper explores how applying Super-Resolution to low-resolution images before Multi-View Stereo reconstruction can enhance the quality and completeness of 3D models, especially in textured scenes, despite potential artifacts.

Contribution

It demonstrates that Super-Resolution improves 3D reconstruction quality in both PatchMatch and deep-learning methods, especially for textured scenes.

Findings

Super-Resolution enhances 3D model completeness.

Improves reconstruction quality in low-resolution scenarios.

Most effective in textured scene reconstruction.

Abstract

Today, Multi-View Stereo techniques are able to reconstruct robust and detailed 3D models, especially when starting from high-resolution images. However, there are cases in which the resolution of input images is relatively low, for instance, when dealing with old photos, or when hardware constrains the amount of data that can be acquired. In this paper, we investigate if, how, and how much increasing the resolution of such input images through Super-Resolution techniques reflects in quality improvements of the reconstructed 3D models, despite the artifacts that sometimes this may generate. We show that applying a Super-Resolution step before recovering the depth maps in most cases leads to a better 3D model both in the case of PatchMatch-based and deep-learning-based algorithms. The use of Super-Resolution improves especially the completeness of reconstructed models and turns out to be particularly effective in the case of textured scenes.