[Paper Review] Robust Watertight Manifold Surface Generation Method for ShapeNet Models

Jingwei Huang, Hao Su+1 more|arXiv (Cornell University)|Feb 5, 2018

Computer Graphics and Visualization TechniquesComputer Science1 references80 citations

TL;DR

The paper presents a robust octree-based isosurface extraction method to convert ShapeNet triangle meshes into watertight 2-manifolds, followed by vertex projection onto the original mesh for precision.

ABSTRACT

In this paper, we describe a robust algorithm for 2-Manifold generation of various kinds of ShapeNet Models. The input of our pipeline is a triangle mesh, with a set of vertices and triangular faces. The output of our pipeline is a 2-Manifold with vertices roughly uniformly distributed on the geometry surface. Our algorithm uses an octree to represent the original mesh, and construct the surface by isosurface extraction. Finally, we project the vertices to the original mesh to achieve high precision. As a result, our method can be adopted efficiently to all ShapeNet models with the guarantee of correct 2-Manifold topology.

Motivation & Objective

• Convert input triangle meshes from ShapeNet into watertight 2-manifolds with uniform surface distribution of vertices.
• Use an octree to efficiently represent the mesh and guide isosurface extraction.
• Ensure final topology is correct and minimize non-manifold issues.
• Provide a post-processing projection step to align the extracted surface with the original geometry.

Proposed method

• Represent the input mesh as a signed distance field using a coarse-to-fine octree with final resolution 0.01.
• Extract the isosurface between positive (outside) and occupied (inside) voxels similar to marching cubes.
• Build octree by subdividing occupied nodes until final resolution is reached.
• Construct connections between neighboring occupied and empty nodes to enable surface extraction.
• Resolve non-manifold ambiguities by splitting vertices/edges where needed during extraction.
• Project extracted vertices back to the original mesh by moving along normals with small steps and apply Laplacian smoothing after each step.

Experimental results

Research questions

• RQ1 Can a robust octree-based isosurface extraction produce watertight 2-manifolds for a wide range of ShapeNet models?
• RQ2 How can the method ensure correct 2-manifold topology across meshes with holes or thin structures?
• RQ3 What is the impact of projecting extracted vertices back to the original mesh on geometric precision?
• RQ4 What fraction of models exhibit face flips, and what causes these flips?

Key findings

• The method yields 2-manifold outputs for all ShapeNet models tested.
• Approximately 5% of models exhibit face flips, typically due to thin original faces.
• The approach guarantees a 2-manifold topology while maintaining high geometric fidelity via projection to the input mesh.