Dr. KID: Direct Remeshing and K-set Isometric Decomposition for Scalable Physicalization of Organic Shapes

Dawar Khan,Ciril Bohak,Ivan Viola

2023-07-24

Abstract:Dr. KID is an algorithm that uses isometric decomposition for the physicalization of potato-shaped organic models in a puzzle fashion. The algorithm begins with creating a simple, regular triangular surface mesh of organic shapes, followed by iterative k-means clustering and remeshing. For clustering, we need similarity between triangles (segments) which is defined as a distance function. The distance function maps each triangle's shape to a single point in the virtual 3D space. Thus, the distance between the triangles indicates their degree of dissimilarity. K-means clustering uses this distance and sorts of segments into k classes. After this, remeshing is applied to minimize the distance between triangles within the same cluster by making their shapes identical. Clustering and remeshing are repeated until the distance between triangles in the same cluster reaches an acceptable threshold. We adopt a curvature-aware strategy to determine the surface thickness and finalize puzzle pieces for 3D printing. Identical hinges and holes are created for assembling the puzzle components. For smoother outcomes, we use triangle subdivision along with curvature-aware clustering, generating curved triangular patches for 3D printing. Our algorithm was evaluated using various models, and the 3D-printed results were analyzed. Findings indicate that our algorithm performs reliably on target organic shapes with minimal loss of input geometry.

Graphics,Computer Vision and Pattern Recognition,Computer Science and Game Theory

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: how to effectively decompose the surfaces of complex organic shapes (such as viruses, organelles or cells) into a finite number of reusable triangular patches, so as to generate 3D jigsaw components suitable for large - scale production. Specifically, the author aims to physicalize these shapes in a low - cost and high - fidelity manner through isometric decomposition and direct remeshing, while ensuring that the finally generated 3D jigsaw components can be easily assembled and disassembled.

### The core challenges of the problem include:
1. **Geometric fidelity**: While simplifying the model, maintain the geometric features of the original biological structure.
2. **Curvature - aware modeling**: Considering the complex curvatures of biological structures, ensure that the decomposed components can truly reflect the original shape.
3. **Minimization of the number of components**: In order to adapt to production methods such as injection molding, it is necessary to minimize the number of different types of components as much as possible.
4. **Smoothness and thickness treatment**: Ensure that the generated 3D - printed components have appropriate thickness and smoothness to achieve a good assembly effect.

### Overview of the solution:
The paper proposes an algorithm named Dr. KID, which achieves the above goals through the following steps:

1. **Initial mesh generation and optimization**: Use the Marching Cubes algorithm to generate an initial mesh from the segmented volume data, and optimize the mesh quality through CVT (Centroidal Voronoi Tessellation).
2. **K - means clustering**: Map triangles to a virtual 3D space according to their side lengths, and use the K - means algorithm for clustering. The goal of clustering is to minimize the distance between triangles within the same category.
3. **Remeshing**: Adjust the mesh through local operations (such as edge flipping, edge folding and vertex translation) to make the triangles within the same category as similar as possible.
4. **Curvature - aware thickening and connector design**: Add appropriate thickness to each triangular patch, and design hinges and holes for assembly.

### Mathematical formulas:
- **Energy function**: The energy function used to measure the clustering effect is defined as follows:
\[
F_d(M_f)=\sum_{i = 1}^{k}\sum_{j = 1}^{n}d(t_j,t_i^*)
\]
where \(k\) is the number of clusters, \(n\) is the total number of triangles, \(c_i\) is the \(i\)-th cluster, \(t_j\) is a triangle belonging to \(c_i\), and \(t_i^*\) is the centroid of \(c_i\).

- **Triangle similarity distance**: The similarity distance between two triangles \(t_j\) and \(t_i^*\) is defined as:
\[
d(t_j,t_i^*) = |x_j - x_i^*|^2+|y_j - y_i^*|^2+|z_j - z_i^*|^2
\]
where \((x_j,y_j,z_j)\) and \((x_i^*,y_i^*,z_i^*)\) respectively represent the coordinates of the two triangles in the virtual 3D space.

Through these steps, the Dr. KID algorithm can generate 3D jigsaw components suitable for large - scale production while maintaining geometric fidelity, thus providing a new solution for the physicalization of biological structures.