A multi-spectral geometric approach for shape analysis
Minimal surfaces in differential geometry, surfaces with prescribed mean curvature (53A10) Applications of differential geometry to data and computer science (53Z50) Spectral problems; spectral geometry; scattering theory on manifolds (58J50) Spectral, collocation and related methods for boundary value problems involving PDEs (65N35) Computer graphics; computational geometry (digital and algorithmic aspects) (68U05)
In this paper the authors introduce a very interesting idea which they show improves numerous tasks in geometric processing such as shape retrieval and similarity between non-rigid objects. The main idea in the current paper is the following. It is known that many solid objects in \(\mathbb R^3\) can be represented by a smooth boundary surface which can be equipped with an intrinsic metric to form a 2-Riemannian manifold. Given that, a way to analyze and process many solid objects in 3D is to treat their shapes as 2-Riemannian manifolds describing their boundary surfaces. The key observation is that the metric chosen for such a surface determines fundamental properties of the shapes by which similarity and other quantities can be evaluated. For example, while two spheres with different radii are non-isometric when equipped with a regular Riemannian metric induced from \(\mathbb R^3\), they become isometric when the 2-manifold is equipped with a scale-invariant metric. This means that two surfaces can be isometrically equivalent with respect to one metric and isometrically different with respect to another.\N\NThis motivates the following: The authors study surfaces using multiple metrics that give birth to multi-spectra by which a given surface can be characterized. Their relative sensitivity to different types of local structures allows each metric to provide a distinct perspective of the shape which then allows for different geometric tasks, say non-rigid identification. The authors achieve their multi-metric approach by considering the spectral decomposition of Laplace-Beltrami operators (LBOs) defined with various metric tensors extending the approach introduced in [\textit{D. Bensaïd} et al., ``Partial matching of non-rigid shapes by learning piecewise smooth functions, Comput. Graph. Forum 42, No. 5, Article ID e14913, 10 p. (2023; \url{doi:10.1111/cgf.14913})] which promoted the concatenation of the spectra of the LBO based on the regular metric and of its scale-invariant version to characterize surfaces and match them to their parts.\N\NThe paper is well written with an excellent set of references.
- A review of trust region algorithms for optimization
- Can One Hear the Shape of a Drum?
- Computing Discrete Minimal Surfaces and Their Conjugates
- scientific article; zbMATH DE number 176057 (Why is no real title available?)
- Introduction to quantum mechanics
- Numerical Geometry of Non-Rigid Shapes
- Scale invariant geometry for nonrigid shapes
- Spectral determination of analytic bi-axisymmetric plane domains
This page was built for publication: A multi-spectral geometric approach for shape analysis
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7009584)
