We introduce Point2Skeleton, an unsupervised method to learn skeletal epresentations from point clouds.
Welcome to check out the paper Point2Skeleton: Learning Skeletal Representations from Point Clouds
The full code is coming soon!
We introduce a generalized skeletal representation, called skeletal mesh. Several good properties of the skeletal mesh make it a useful representation for shape analysis:
Recoverability The skeletal mesh can be considered as a complete shape descriptor, which means it can reconstruct the shape of the original domain.
Abstraction The skeletal mesh captures the fundamental geometry of a 3D shape and extracts its global topology; the tubular parts are abstracted by simple 1D curve segments and the planar or bulky parts by 2D surface triangles.
Structure awareness The 1D curve segments and 2D surface sheets as well as the non-manifold branches on the skeletal mesh give a structural differentiation of a shape.
Volume-based closure The interpolation of the skeletal spheres gives solid cone-like or slab-like primitives; then a local geometry is represented by volumetric parts, which provides better integrity of shape context. The interpolation also forms a closed watertight surface.
The skeletal mesh provides a suitable vehicle for solving this problem, since we can reconstruct the surfaces of the input point clouds by interpolating the skeletal spheres. On the one hand, the reconstructions using skeletal meshes preserve the complex typologies and also capture the thin structures of the input. On the other hand, unlike the Poisson reconstruction, our method does not need to input any normal information, and is still able to produce high-quality watertight surfaces.
Unsupervised Structural Decomposition
The skeletal mesh is structure-aware; thus it naturally induces a structural decomposition of a shape without data annotation by the non-manifold branches and dimensional changes (curve-surface joints).