Learning Quadrangulated Patches For 3D Shape Processing

TL;DR

This paper introduces a novel 3D surface inpainting method using local patches derived from mesh quadrangulation, employing generative models to effectively repair moderate-sized holes in 3D shapes.

Contribution

It presents a new encoding of local patches via height maps based on mesh quadrangulation and compares dictionary and autoencoder models for improved surface inpainting.

Findings

Autoencoder models outperform previous geometry-based methods.

Method works on both synthetic and real-world 3D shapes.

Effective repair of moderate-sized holes in 3D meshes.

Abstract

We propose a system for surface completion and inpainting of 3D shapes using generative models, learnt on local patches. Our method uses a novel encoding of height map based local patches parameterized using 3D mesh quadrangulation of the low resolution input shape. This provides us sufficient amount of local 3D patches to learn a generative model for the task of repairing moderate sized holes. Following the ideas from the recent progress in 2D inpainting, we investigated both linear dictionary based model and convolutional denoising autoencoders based model for the task for inpainting, and show our results to be better than the previous geometry based method of surface inpainting. We validate our method on both synthetic shapes and real world scans.