Abstract:We present a novel shape-approximating anisotropic re-meshing algorithm as a geometric generalization of the adaptive moving mesh method. Conventional moving mesh methods reduce the interpolation error of a mesh that discretizes a given function over a planar domain. Our algorithm, in contrast, optimizes the mesh's approximation of a curved surface; surfaces can be represented in various formats, such as a signed distance field. The optimization is achieved by continuously flowing the mesh without altering its topology, making the implementation simpler compared to other adaptive surface meshing techniques. The resulting optimal mesh can be interpreted as a harmonic map with respect to a metric using the shape operator. Furthermore, our approach can be tailored to target height fields by utilizing isotropic geometry.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is how to find a triangulation of an arbitrary surface that is both accurate and concise in computer graphics and computer - aided design. Specifically, the paper introduces a new method that can iteratively evolve an existing mesh to approximate the ideal surface. Unlike previous methods, this iterative process does not require complex operations to change the topology of the mesh. The main contribution of the paper lies in generalizing the moving - mesh method from function interpolation optimization on a planar domain to surface approximation and providing a method to optimize the mesh while keeping the mesh topology unchanged.
### Core problems of the paper
1. **Surface approximation**: How to optimize the approximation of the surface by the mesh while keeping the mesh topology unchanged.
2. **Adaptive remeshing**: How to optimize the approximation of the surface by continuously flowing the mesh without changing the topology of the mesh.
3. **Geometric generalization**: How to extend the traditional moving - mesh method from function interpolation optimization on a planar domain to surface approximation.
### Specific problems
- **Improvement of the initial mesh**: Given an initial mesh, how to optimize its approximation of the target shape through continuous heat - diffusion - like flows.
- **Description of the mesh flow**: How to use the metric defined by the shape operator to minimize the Dirichlet energy, thereby obtaining a harmonic mapping.
- **Anisotropic control**: How to control the anisotropy of the generated triangles by introducing anisotropic parameters.
### Solutions
- **Harmonic mapping**: By minimizing the Dirichlet energy, find a harmonic mapping so that the mesh can approximate the given shape efficiently and adaptively while keeping the topology unchanged.
- **Anisotropic optimization**: By introducing anisotropic parameters, control the anisotropy of the generated triangles, so as to better adapt to different regions of the surface.
- **Numerical implementation**: Through discretizing the energy function and the gradient - descent method, achieve the specific numerical solution of the algorithm.
### Main contributions
1. **Geometric generalization**: Extend the moving - mesh theory from the planar domain to the surface.
2. **Surface - approximation application**: Apply the generalized moving - mesh method to the surface - approximation problem, only requiring the provision of an SDF and an initial mesh.
3. **Implementation details**: Describe in detail the anisotropic control, boundary treatment, and the method for calculating the shape operator of each face.
### Mathematical formulas
- **Dirichlet energy**:
\[
E(\phi):=\frac{1}{2}\int_M\sigma(d\phi\wedge\star_g d\phi)
\]
where \(\sigma\) is the metric on the target manifold \(M\), \(g\) is the Riemannian metric on the material space \(cM\), and \(\phi: cM\rightarrow M\) is the mapping from the material space to the target manifold.
- **Anisotropic energy**:
\[
E_{2D,\alpha}(\phi)=\int_{cM}h(|S|^\alpha d\phi\wedge\star|S|^\alpha d\phi)
\]
where \(\alpha\in[0, 1]\) is the anisotropic parameter, and \(|S|\) is the absolute value of the shape operator.
- **Discrete energy**:
\[
E_{2D}|_i(\phi)=\sum_{j\in\text{Star}(i)}\||S|d\phi_{ij}\|^2
\]
where \(d\phi_{ij}\) represents the discrete differential along the edge \(ij\).
Through these methods and formulas, the paper provides an effective and flexible solution for optimizing the triangulation of the surface.