How many points in a point cloud is sufficient for accurate estimation of the curvature
R. Mirzaie · 2025
Read How many points in a point cloud is sufficient for accurate estimation of the curvature by R. Mirzaie (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-06779.txt.
Introduction
Mirzaie asks, in 2025, how many sampled points are needed to estimate the curvature of a curve or surface accurately. Point clouds are finite samples drawn from some distribution spread over the shape. Too few points near a spot make local bending impossible to judge.
An algorithm first estimates curvature at a point of a curve from its nearby samples. The method then extends to the Gaussian curvature of surfaces. A key relation ties the number of selected points to the chance that a given point has enough neighbors. That relation lets users control how large the cloud must be before measuring. The work serves scanning and shape analysis.
Catalog record
| CREATOR | R. Mirzaie |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 26 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-06779
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
How many points in a point cloud is sufficient for accurate estimation of the curvature Thanks: R. Mirzaie, Department of Pure Mathematics, Faculty of Science, Imam Khomeini International University, Qazvin, Iran [email protected] R. Mirzaie Abstract . We introduce an estimator for the curvature of curves and surfaces by using finite sample points drawn from sampling a probability distribution that has support on the curve or surface. First we give an algorithm for estimation of the curvature in a given point of a curve. Then, we extend it to estimate the Gaussian curvature of the surfaces. In the proposed algorithms, we use a relation between the number of selected points in the point cloud and the probability that a given point has a sufficient number of nearby points. This relation allows us to control the required number of points in the point cloud. Key words: Curve, Surface, Curvature, Manifold learning. MSC: 53C21, 53C23, 62D05. 1. Introduction Recently, there has been significant progress in the manifold learning and has been attractive for many researches, because of the useful applications in topological data analysis, geometric measure theory, integral geometry, image processing arguments and biomathematics. But, a main problem is still open: What is the best algorithm for estimating the curvature of a manifold using a given point cloud. There are many algorithms suited for Gaussian curvature and mean curvature estimation. Almost all of the algorithms rely on the estimating of intrinsic geometric properties of a surface from a polygonal mesh obtained from data. For example, a widely used approximation of the curvature of a surface is based on the fact that a surface around a point can be represented as a graph of a bivariate function, and the curvature can be approximated to the curvature of a hyperbola passing from the neighbouring points of the given point in the selected point cloud( see [12]). The authors of [1] use principal component analysis to estimate the Ricci curvature of a manifold at each point of a point cloud set, selected randomly from the points of a Riemannian submanifold of the euclidean space. Estimator of [9], for the scalar curvature, is based on the fact that the scalar curvature at a point x of a Riemannian manifold characterizes the growth rate of the volume of a geodesic ball B(x,r) as r decreases. Thus, estimation of the scalar curvature reduces to estimation of the volumes. We refer to [ 3, 4, 5, 6, 8, 9, 14] for many methods to estimate Gaussian curvature from point clouds that are sampled from surfaces. Also, we refer the reader to [2, 10, 12, 13] for surveys. The authors of [7], locally fit a set of circles through the neighbour vertices of the triangulations, as a main tool for estimations. In [2] a suitable curve is constructed at each vertex of the surface triangulation and then the local differential structure is derived. In the present article, we give an algorithm to estimate the curvature of a curve by using the osculating circle at a given point. Then, we use it to estimate the Gussian curvature of surfaces. Using osculating circle is a traditional method for estimation of the curve of plane curves, but using it for estimation of the principal curvatures of the surfaces from random point clouds is new. Another subject which is new in the present article is the following fact: Using random point clouds to estimate geometric properties with global characteristics works very well. But, when we use random point clouds to estimate geometric properties with local characteristics (for example curvature at a point), there may be significant estimation errors. Let a be an arbitrary point on a surface M and B=\{x_{1},...,x_{m}\} be a subset of points on M . To estimate the curvature at the point a , the points of B that are close to a play essential role. But, we are uncertain if there are enough points in B close to a . For example, in the following picture we have chosen points randomly on the surface. The selected points are not suitable for estimation of the curvature at b . But, they are suitable for the curvature estimation at a In practice, since we usually have no information about the underlying manifold, the easiest way to solve the problem is to choose a large number of points. But, the other problem is that the large number of points makes the computation process and runtime of the computer program long. To overcome these two problems, we use Theorem 2.7 and Theorem 2.14, to provide a method for finding the minimum number m for the points in the point cloud. m is suitable with a probability p , where p can be chosen arbitrary close to one. 2. Results In what follows, each curve and surface is considered to be differentiable with the finite length and area imbedded in R^{2} and R^{3} , respectively. Since we intend to estimate their curvature at a generic point, we suppose that they have not boundary points. We suppose that each sample set B=\{x_{1},...,x_{m}\} of the points on the curve or surface is i.i.d. (independent and identically distributed) uniformly distributed ( uniform with respect to the Riemannian volume measure induced from the ambient space). Remark 2.1. Let \gamma be a differentiable plan curve and let r=r(a,b,c) be radius of the circle generated by three nonlinear points a,b,c belonging to \gamma . The curvature of \gamma at the point a which we denote by \kappa(a) is defined by to \kappa(a)^{-1}=lim_{{}_{(b,c)\to(a,a)}}r(a,b,c) . A circle with radius \kappa(a)^{-1} tangent to \gamma at the point a is called the osculating circle at a . Thus, for the points b,c sufficiently close to a , \frac{1}{r} is an estimation for \kappa(a) . If max\{|b-a|,|c-a|\}<\epsilon , then we say that \frac{1}{r} is a \epsilon -triangle estimation of \kappa(a) . Let d_{1},d_{2},d_{3} be the length of the sides of the triangle abc and D=\frac{d_{1}+d_{2}+d_{3}}{2} . Then, r=\frac{d_{1}d_{2}d_{3}}{4\sqrt{D(D-d_{1})(D-d_{2})(D-d_{3})}}. Thus, \epsilon -triangle estimation of \kappa(a) is \kappa(a)=\frac{4\sqrt{D(D-d_{1})(D-d_{2})(D-d_{3})}}{d_{1}d_{2}d_{3}}. In what follows, we denote by d(x,y) the distance between x,y in R^{n} . If S is a connected curve or a surface, then the distance between x,y in S is denoted by d_{S}(x,y) , which is equal to the infimum of the length of the curves in S joining x to y . We denote by \gamma and M a differentiable plan curve with the finite length and a differentiable surface with the finite area, respectively. B(x,r) is a ball with the center x and of radius r>0 ( in R^{n} ). B(r) will denote a ball of radius r ( non-specified center). If M is a surface in R^{3} and x\in M , we denote by B_{M}(x,r) the geodesic ball in M with the center x and the radius r , that is B_{M}(x,r)=\{y\in M:d_{M}(x,y)\leq r\}. We use the symbol vol(M) to denote the volume(area) of M . Remark 2.2. vol(B_{M}(x,r)) is a continuous function of r . Thus, for each positive number \epsilon ( sufficiently small ), there is a number r=r(x,\epsilon) such that vol(B_{M}(x,r))=\pi\epsilon^{2} ( the area of the usual ball with the radius \epsilon in a plane). In this case, we denote B_{M}(x,r) by B^{\prime}(x,\epsilon) .