Geometry Enhancements from Visual Content: Going Beyond Ground Truth

TL;DR

This paper introduces a cyclic architecture that extracts and reintegrates high-frequency image patterns to improve depth sensor resolution and generate detailed, realistic 3D models, surpassing previous methods.

Contribution

It proposes a novel cyclic framework for enhancing depth resolution and 3D model quality by leveraging high-frequency visual content extraction and reinsertion.

Findings

Achieves state-of-the-art depth super-resolution results

Produces visually attractive, detailed 3D models

Enhances low-cost depth sensor outputs

Abstract

This work presents a new cyclic architecture that extracts high-frequency patterns from images and re-insert them as geometric features. This procedure allows us to enhance the resolution of low-cost depth sensors capturing fine details on the one hand and being loyal to the scanned ground truth on the other. We present state-of-the-art results for depth super-resolution tasks and as well as visually attractive, enhanced generated 3D models.