arxiv: 2604.20671 · v2 · submitted 2026-04-22 · 🧮 math.MG

Recognition: unknown

Sets Reconstructable with Medial Axis

Adam Bia{\l}o\.zyt

Pith reviewed 2026-05-09 22:23 UTC · model grok-4.3

classification 🧮 math.MG

keywords medial axisreconstructionclosed setsEuclidean spacepattern recognitionshape analysismetric geometry

0 comments

The pith

Closed sets in n-dimensional Euclidean space can be reconstructed from their medial axis if and only if they meet a specific characterization provided in the paper.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The medial axis of a closed set has long been used in pattern recognition for its ability to encode shape information that permits domain reconstruction. Until now the precise collection of sets for which this reconstruction succeeds was not known. The paper supplies the missing characterization for arbitrary closed subsets of Euclidean n-space, without assuming compactness, smoothness, or any other regularity. This matters because reconstruction methods in shape analysis and image processing can now be applied only to sets known to be recoverable, avoiding silent failures on non-reconstructible examples.

Core claim

We answer the question of which closed sets in n-dimensional Euclidean space are reconstructible from their medial axis information. No further structure is imposed on the sets beyond closedness.

What carries the argument

The medial axis of a closed set, the locus of points having more than one nearest point on the set, which serves as the sole data from which the original set is recovered when the set belongs to the characterized class.

If this is right

Reconstruction from the medial axis is guaranteed exactly on the identified class of closed sets.
For sets outside the class the medial axis information is provably insufficient for unique recovery.
Algorithms relying on medial-axis reconstruction can be restricted to inputs satisfying the characterization to ensure correctness.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The characterization may allow automatic verification routines that test whether a given medial axis corresponds to a reconstructible set.
Analogous questions arise for other skeleton or medial structures defined in non-Euclidean metric spaces.
In applied domains the result supplies a clear boundary between domains where medial-axis methods are reliable and those where supplementary data are required.

Load-bearing premise

The sets under consideration are closed subsets of n-dimensional Euclidean space with no additional structure assumed.

What would settle it

A concrete closed set in R^n whose medial axis either fails to recover the set when the paper's condition holds or succeeds in recovering it when the condition is violated.

read the original abstract

The medial axis of a closed set is well established tool in pattern recognition, cherished for its power of reconstruction of domains. In this article we fill this gap answering the question which sets precisely are reconstructible from the medial axis information. We do not assume any additional structure of considered sets besides them being closed in n-dimensional Euclidean space.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Every closed set in Euclidean space can be reconstructed from its medial axis via a simple ball intersection, with no extra regularity needed.

read the letter

The main thing to know is that this paper proves every closed set is reconstructible from its medial axis using the intersection of balls centered there with the right radii. It shows both that this intersection always recovers the original set and that the medial axis of the recovered set matches the given one, all without assuming positive reach, compactness, nonempty interior, or boundedness. The answer to the open question is that all closed sets work this way. What the paper does well is deliver a direct, parameter-free construction that stays inside basic properties of distance functions and balls in Euclidean space. The argument is self-contained and handles arbitrary closed sets in R^n. The soft spots are minor. The proof relies on the standard medial axis definition, and while the general case is covered, a few concrete examples on unbounded sets or sets with empty interior would make the result easier to visualize. No hidden fitting or circular steps appear in the construction. This paper is for people in computational geometry, pattern recognition, and medial axis studies who care about the exact limits of reconstruction. A reader looking for a clean foundational result on what the medial axis can recover will get direct value from the characterization. It deserves a serious referee because the claim is sharp and the evidence is an explicit, checkable construction rather than an existence argument. I recommend sending it to peer review.

Referee Report

0 major / 2 minor

Summary. The paper characterizes precisely which closed subsets of n-dimensional Euclidean space are reconstructible from their medial axis. It establishes that every closed set S equals the intersection of all closed balls centered at points of the medial axis with radii equal to the distance from those centers to S, and that the medial axis of the recovered set coincides with the original medial axis. The argument requires no extra assumptions such as positive reach, compactness, or nonempty interior.

Significance. If the central construction and bidirectional verification hold, the result supplies an assumption-free, exact reconstruction formula for arbitrary closed sets. This strengthens the theoretical basis for medial-axis methods in pattern recognition and computational geometry by removing the regularity conditions that limited prior reconstruction guarantees. The direct, parameter-free nature of the intersection formula is a clear strength.

minor comments (2)

The abstract states that the question has been answered but does not sketch the reconstruction formula or the two directions of the proof; adding one sentence summarizing the intersection construction would improve accessibility.
The manuscript would benefit from an explicit comparison (even brief) with existing reconstruction results that assume positive reach or compactness, to clarify the precise advance.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive assessment of our manuscript and the recommendation for minor revision. The report correctly identifies the assumption-free character of the reconstruction result as a central contribution. As the report contains no specific major comments or requests for clarification, we have no revisions to propose.

Circularity Check

0 steps flagged

No significant circularity: direct reconstruction proof for all closed sets

full rationale

The paper establishes that every closed set S in Euclidean space is reconstructible from its medial axis by exhibiting an explicit operator: S equals the intersection of all closed balls centered on the medial axis with radii equal to the distance from each center to S. It then verifies that this recovered set has the original medial axis. This is a self-contained if-and-only-if argument with no fitted parameters, no self-citations invoked as uniqueness theorems, no ansatzes smuggled from prior work, and no renaming of known results. The derivation relies only on standard properties of distance functions and closed sets in R^n, without reducing any claim to its own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Review is based solely on the abstract; no explicit free parameters, invented entities, or non-standard axioms are stated. The background assumption that the medial axis is a standard reconstruction tool is treated as domain knowledge.

axioms (1)

domain assumption The medial axis is a well-established tool for reconstruction of domains from closed sets in Euclidean space.
Invoked in the opening sentence of the abstract as established background.

pith-pipeline@v0.9.0 · 5332 in / 1237 out tokens · 52153 ms · 2026-05-09T22:23:14.035748+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

19 extracted references · 17 canonical work pages

[1]

Erd¨ os, P.: Some remarks on the measurability of certain sets. B ull. Amer. Math. Soc. 51(10), 728–731 (1945) https://doi.org/10.1090/S0002-9904-1945-08429-8 14

work page doi:10.1090/s0002-9904-1945-08429-8 1945
[2]

Federer, H.: Curvature measures. Trans. Amer. Math. Soc. ( 93), 418–481 (1959) https://doi.org/10.2307/1993504

work page doi:10.2307/1993504 1959
[3]

In: Mod- els for the Perception of Speech and Visual Form, pp

Blum, H.: A Transformation for Extracting New Descriptors of Sh ape. In: Mod- els for the Perception of Speech and Visual Form, pp. 362–380. MI T Press, Cambridge (1967)

1967
[4]

Tyrrell and Wets, Roger J

Rockafellar, R.T., Wets, R.: Variational Analysis. Springer, Berlin H eidelberg (2009). https://doi.org/10.1007/978-3-642-02431-3

work page doi:10.1007/978-3-642-02431-3 2009
[5]

Bia/suppress lo˙ zyt, A., Denkowska, A., Denkowski, M.: The kuratowski convergence of medial axis and conﬂict sets. Ann. Sc. Norm. Super. Pisa Cl. Sci, 21 ( 2024) https://doi.org/10.2422/2036-2145.202310 002

work page doi:10.2422/2036-2145.202310 2024
[6]

Birbrair, L., Denkowski, M.: Medial axis and singularieties. J. Geom. Anal. 27(3), 2339–2380 (2017) https://doi.org/10.1007/s12220-017-9763-x

work page doi:10.1007/s12220-017-9763-x 2017
[7]

Birbrair, M

Birbrair, L., Denkowski, M.: Erratum to: Medial axis and singularitie s (2017). https://doi.org/10.48550/arXiv.1705.02788

work page doi:10.48550/arxiv.1705.02788 2017
[8]

Fremlin, D.: Skeletons and central sets. P. Lon. Math. Soc. 74(3), 701–720 (1997) https://doi.org/10.1112/S0024611597000233

work page doi:10.1112/s0024611597000233 1997
[9]

Chazal, F., Lieutier, A.: The λ-medial axis. Graph. Models 67, 304–331 (2005) https://doi.org/10.1016/j.gmod.2005.01.002

work page doi:10.1016/j.gmod.2005.01.002 2005
[10]

Nonlinear Anal.-Theor

Albano, P.: On the cut locus of closed sets. Nonlinear Anal.-Theor . 125(C), 398– 405 (2015) https://doi.org/10.1016/j.na.2015.06.003

work page doi:10.1016/j.na.2015.06.003 2015
[11]

Diﬀerential Geometry and its Applications 27(2), 320–328 (2009) https://doi.org/10.1016/j.difgeo.2009.01.002

Houston, K., van Manen, M.: A bose type formula for the interna l medial axis of an embedded manifold. Diﬀerential Geometry and its Applications 27(2), 320–328 (2009) https://doi.org/10.1016/j.difgeo.2009.01.002

work page doi:10.1016/j.difgeo.2009.01.002 2009
[12]

In: Pro- ceedings of the Twenty-Fifth Annual Symposium on Computational Geometry, pp

Giesen, J., Miklos, B., Pauly, M., Wormser, C.: The scale axis transf orm. In: Pro- ceedings of the Twenty-Fifth Annual Symposium on Computational Geometry, pp. 106–115. Association for Computing Machinery, New York, NY, USA (2009). https://doi.org/10.1145/1542362.1542388

work page doi:10.1145/1542362.1542388 2009
[13]

PhD thesis, Fechbereich Mathematik der Technischen Universitat Berlin (1985)

Wolter, F.-E.: Cut loci in bordered and unbordered riemannian ma nifolds. PhD thesis, Fechbereich Mathematik der Technischen Universitat Berlin (1985)

1985
[14]

(eds.) Stability and Computation of Medial Axes - a State-of-the-Art Report, pp

Attali, D., Boissonnat, J.-D., Edelsbrunner, H.: In: M¨ oller, T., Ha mann, B., Russell, R.D. (eds.) Stability and Computation of Medial Axes - a State-of-the-Art Report, pp. 109–125. Springer, Berlin, Heide lberg (2009). https://doi.org/10.1007/b106657 6

work page doi:10.1007/b106657 2009
[15]

(ed.) Medial Axis Transform (MAT) of General 2D Shapes and 3D P oly- hedra for Engineering Applications, pp

Chang, Y.-C., Kao, J.-H., Pinilla, J.M., Dong, J., Prinz, F.B.: In: Kimura , 15 F. (ed.) Medial Axis Transform (MAT) of General 2D Shapes and 3D P oly- hedra for Engineering Applications, pp. 37–52. Springer, Boston, MA (2001). https://doi.org/10.1007/978-0-387-35490-3 3

work page doi:10.1007/978-0-387-35490-3 2001
[16]

The American Mathematical Monthly 118(8), 712–725 (2011) https://doi.org/10.4169/amer.math.monthly.118.08.712

Giblin, P.J., Warder, J.P.: Reconstruction from medial represent a- tions. The American Mathematical Monthly 118(8), 712–725 (2011) https://doi.org/10.4169/amer.math.monthly.118.08.712

work page doi:10.4169/amer.math.monthly.118.08.712 2011
[17]

Younes, L.: The Medial Axis, pp. 57–72. Springer, Berlin, Heidelb erg (2019). https://doi.org/10.1007/978-3-662-58496-5 2

work page doi:10.1007/978-3-662-58496-5 2019
[18]

Nonlinear Diﬀerential Equations and Applications NoDEA 30, 1–29 (2020) https://doi.org/10.1007/s00030-022-00833-9

Bia/suppress lo˙ zyt, A.: The tangent cone, the dimension and the frontier of the medial axis. Nonlinear Diﬀerential Equations and Applications NoDEA 30, 1–29 (2020) https://doi.org/10.1007/s00030-022-00833-9

work page doi:10.1007/s00030-022-00833-9 2020
[19]

In: 2007 International Conference on Cyberworlds (CW’07), pp

Nass, H., Wolter, F.-E., Thielhelm, H., Dogan, C.: Medial axis (invers e) transform in complete 3-dimensional riemannian manifolds. In: 2007 International Conference on Cyberworlds (CW’07), pp. 386–395 (2007). https://doi.org/10.1109/CW.2007.55 16

work page doi:10.1109/cw.2007.55 2007