Neural Geometry Image-Based Representations with Optimal Transport (OT)

TL;DR

This paper introduces a novel neural geometry image-based representation for 3D meshes that leverages optimal transport to enable efficient, decoder-free surface restoration with high quality and storage efficiency.

Contribution

It proposes a geometry image-based representation that transforms irregular meshes into regular images, enabling efficient neural processing and continuous levels of detail using optimal transport.

Findings

Achieves state-of-the-art compression ratio and restoration accuracy.

Supports single-pass high-quality mesh reconstruction.

Demonstrates superior storage efficiency and detail preservation.

Abstract

Neural representations for 3D meshes are emerging as an effective solution for compact storage and efficient processing. Existing methods often rely on neural overfitting, where a coarse mesh is stored and progressively refined through multiple decoder networks. While this can restore high-quality surfaces, it is computationally expensive due to successive decoding passes and the irregular structure of mesh data. In contrast, images have a regular structure that enables powerful super-resolution and restoration frameworks, but applying these advantages to meshes is difficult because their irregular connectivity demands complex encoder-decoder architectures. Our key insight is that a geometry image-based representation transforms irregular meshes into a regular image grid, making efficient image-based neural processing directly applicable. Building on this idea, we introduce our neural geometry image-based representation, which is decoder-free, storage-efficient, and naturally suited for neural processing. It stores a low-resolution geometry-image mipmap of the surface, from which high-quality meshes are restored in a single forward pass. To construct geometry images, we leverage Optimal Transport (OT), which resolves oversampling in flat regions and undersampling in feature-rich regions, and enables continuous levels of detail (LoD) through geometry-image mipmapping. Experimental results demonstrate state-of-the-art storage efficiency and restoration accuracy, measured by compression ratio (CR), Chamfer distance (CD), and Hausdorff distance (HD).