Geometry-Aware Representation Denoising for Robust Multi-view 3D Reconstruction

2026-05-25

AI Summary

This work addresses the limited robustness of multi-view 3D reconstruction under real-world image degradation by introducing GARD, a novel framework that, for the first time, integrates diffusion-based denoising into the geometry-aware feature space of a feed-forward 3D reconstruction model. By jointly optimizing geometric structure and RGB image quality at the feature level, GARD achieves unified high-fidelity 3D reconstruction and image restoration. The method combines diffusion modeling, multi-view feature denoising, and an RGB decoder to significantly enhance reconstruction robustness and visual fidelity in degraded scenarios, as demonstrated on the Depth Anything 3 benchmark.

Abstract

Multi-view 3D reconstruction has achieved remarkable progress with the advent of feed-forward 3D reconstruction models. However, these models are typically trained and evaluated under ideal, degradation-free imaging conditions, whereas real-world observations often contain degradations that differ significantly from such settings. Improving robustness for multi-view 3D reconstruction under degraded conditions therefore remains an important challenge. We present Geometry-Aware Representation Denoising (GARD), a novel framework that performs diffusion-based multi-view restoration directly in the feature space of a feed-forward 3D reconstruction model. This design exploits the geometry-aware feature representations of the 3D reconstructor to effectively recover accurate scene geometry. Furthermore, by employing an additional RGB image decoder, the refined representations can also be used to restore high-quality RGB images, thereby enabling the simultaneous recovery of 3D scene geometry and high-quality imagery. Comprehensive experiments on the Depth Anything 3 (DA3) benchmark demonstrate the effectiveness of the proposed GARD framework.

Problem

• multi-view 3D reconstruction
• robustness
• degraded conditions
• geometry recovery
• real-world imaging

Innovation

• Geometry-Aware Representation
• Diffusion-based Restoration
• Feature-space Denoising
• Multi-view 3D Reconstruction
• Joint Geometry and Image Recovery