Neural Geometry Processing via Spherical Neural Surfaces

Romy Williamson,Niloy J. Mitra

2024-07-10

Abstract:Neural surfaces (e.g., neural map encoding, deep implicits and neural radiance fields) have recently gained popularity because of their generic structure (e.g., multi-layer perceptron) and easy integration with modern learning-based setups. Traditionally, we have a rich toolbox of geometry processing algorithms designed for polygonal meshes to analyze and operate on surface geometry. However, neural representations are typically discretized and converted into a mesh, before applying any geometry processing algorithm. This is unsatisfactory and, as we demonstrate, unnecessary. In this work, we propose a spherical neural surface representation (a spherical parametrization) for genus-0 surfaces and demonstrate how to compute core geometric operators directly on this representation. Namely, we show how to construct the normals and the first and second fundamental forms of the surface, and how to compute the surface gradient, surface divergence and Laplace Beltrami operator on scalar/vector fields defined on the surface. These operators, in turn, enable us to create geometry processing tools that act directly on the neural representations without any unnecessary meshing. We demonstrate illustrative applications in (neural) spectral analysis, heat flow and mean curvature flow, and our method shows robustness to isometric shape variations. We both propose theoretical formulations and validate their numerical estimates. By systematically linking neural surface representations with classical geometry processing algorithms, we believe this work can become a key ingredient in enabling neural geometry processing.

Graphics,Artificial Intelligence,Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The problem this paper attempts to address is how to perform geometric processing directly on surfaces represented by neural networks (i.e., neural surfaces) without discretizing them into polygonal meshes. Specifically, the paper proposes a new representation method called "Spherical Neural Surfaces" (SNS) for seamlessly encoding surfaces with zero topology type and demonstrates how to compute core geometric operators (such as normals, first and second fundamental forms, curvature, continuous Laplace-Beltrami operator, and its minimal eigenmodes) directly on this representation, avoiding unnecessary discretization errors. In this way, the paper aims to provide a new framework for neural geometric processing, allowing traditionally mesh-based geometric processing algorithms to be directly applied to neural representations, thereby improving accuracy and efficiency. Additionally, the paper demonstrates the robustness and accuracy of this method in applications such as neural spectral analysis, heat flow, and mean curvature flow.