Non-symmetric integral geometry: alternative reconstruction procedure with dual Radon transform

arXiv:2611.07376 · hep-th, math-ph, math.FA, math.MP · Submitted 2026-11-05 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: High Energy Physics: Theory. The day in high-energy theory: gravity, strings, fields and the questions left open.

Kai: We're looking at the paper: "Non-symmetric integral geometry".

Mira: Here's the gist of the paper. In this paper we use principles of non-symmetric integral geometry to extend relations traced from the Fourier slice theorem and dual Radon transform for an alternative reconstruction procedure that impacts image reconstruction processes.

Kai: The derived relations confirm the complexity of the inversion operator due to the symmetry breaking of integration measure.

Mira: Integral geometry is backed by symmetry properties of measures on geometrical spaces, and the direct Radon transform is a main object in integral geometry.

Kai: The inverse Radon transforms are used to reconstruct images, but they often face ill-posedness necessitating regularization.

Paper summary: Kai: So, we're looking at this paper, "Non-symmetric integral geometry: alternative reconstruction procedure with dual Radon transform." The authors are using non-symmetric integral geometry to extend known relations from the Fourier slice theorem and the dual Radon transform.

Mira: Their main point is that this method offers an alternative way to reconstruct the image function, and it directly impacts how we handle image reconstruction processes because they show that the inversion operator is actually more complex than previously thought due to a symmetry breaking in how they integrate things <ref:2611.07376#pg3>.

Lev: So, what does this complexity look like on the ground? The paper points out that when the original function being reconstructed lacks those expected symmetry properties, the Radon angle dependence gets restricted instead of integrating over the full range from zero to two pi <ref:2611.07376#pg3>.

Kai: And because of those restrictions, they found that the universal Radon inversion turns into a more complex transformation involving these two contributions, fS and fA <ref:2611.07376#pg3>.

Mira: Specifically, this term fA is what they identify as being induced by the complex integration measure <ref:2611.07376#pg3>. This complexity confirms that the inversion operator itself is more complicated than standard methods suggest <ref:2611.07376#pg2>.

Lev: From an engineering standpoint, this means when you try to run this on real hardware, these extra terms fS and fA are what introduce the overhead in your computation <ref:2611.07376#pg4>.

Kai: Exactly, so they propose inserting a generalized representation for the Fourier transform of the dual Radon image into the standard inverse Fourier transform to create this new extended reconstruction procedure <ref:2611.07376#pg4>.

Mira: That leads them to derive a new and extended formula for reconstructing the function f from its Fourier data, which incorporates these complexities and provides a new way to handle those ill-posed problems <ref:2611.07376#pg5>.

Lev: I just want to be clear that this paper isn't proposing some simple fix for regularization; it's mapping out the full complexity of the inversion operator because of this symmetry breaking <ref:2611.07376#pg3>. It shows exactly what the new computational overhead would be when you try to implement this on real hardware <ref:2611.07376#pg4>.

Kai: So, we’ve established that this paper introduces a framework for handling these non-symmetric scenarios by looking at dual Radon transforms in Fourier space <ref:2611.07376#pg2>. This sets the stage for what we'll discuss next about how these mathematical concepts actually apply to the actual reconstruction process <ref:2611.07376#pg5>.

Mira: So, to wrap up, this paper "Non-symmetric integral geometry: alternative reconstruction procedure with dual Radon transform" explores extending established relations by using non-symmetric integral geometry principles <ref:2611.07376#pg1>. It shows that the symmetry breaking in the integration measure makes the inversion operator more complex <ref:2611.07376#pg3>.

Kai: The authors propose a new and extended representation for image reconstruction that accounts for these complexities, using a generalized Fourier transform of the dual Radon image <ref:2611.07376#pg4>.

Mira: It’s important to remember that this complexity arises because the standard principles of integral geometry were built on certain symmetry properties <ref:2611.07376#pg3>. When those symmetries break, like with a well-localized function that doesn't have them, you get these extra terms fS and fA <ref:2611.07376#pg3>.

Lev: I just want to be clear that this paper isn't proposing some simple fix; it’s mapping out the full complexity of the inversion operator because of this symmetry breaking <ref:2611.07376#pg3>. It shows exactly what the new computational overhead would be when you try to implement this on real hardware <ref:2611.07376#pg4>.

Kai: So, we’ve established that this paper introduces a framework for handling these non-symmetric scenarios by looking at dual Radon transforms in Fourier space <ref:2611.07376#pg2>. This sets the stage for what we'll discuss next about how these mathematical concepts actually apply to the actual reconstruction process <ref:2611.07376#pg5>.

Conclusion: Kai: So, to wrap up this paper, "Non-symmetric integral geometry: alternative reconstruction procedure with dual Radon transform," they’re showing how using non-symmetric math extends old ideas from the Fourier slice theorem and dual Radon transforms.

Mira: They are focusing on how the way they integrate things—the measure—breaks symmetry. And that breaking of symmetry directly leads to a more complicated inversion operator than we thought we had before.

Lev: From an engineering standpoint, that complexity means you can't just use the old simple formula anymore; you have to deal with these new terms arising from the way they integrate <ref:2611.07376#pg3>.

Kai: Right, so the whole point is they derived a brand new way to reconstruct an image function using this dual Radon approach, which is really more robust when you're dealing with functions that aren't perfectly symmetric.

Mira: It means for problems where your original data isn't smooth or perfectly balanced, this method gives you an alternative reconstruction procedure that actually works better than the standard one.

Lev: But the caveat is, it’s not just a theoretical fix; it’s showing you exactly what kind of extra computational overhead you introduce when running this on actual hardware <ref:2611.07376#pg4>.

Kai: And that complexity comes from how they define the Fourier transform of the dual Radon image, which is basically a more detailed description of the measurement itself <ref:2611.07376#pg4>.

Mira: So, when we look at this title and who wrote it—the authors are pushing the boundaries of what integral geometry can handle by looking at these dual transforms in Fourier space <ref:2611.07376#pg2>.

Lev: It’s a deep dive into the math that shows you where the limitations are when you try to apply these reconstruction techniques to real-world imaging problems.

Kai: And that leads us into how these mathematical concepts actually play out in the actual image reconstruction process <ref:2611.07376#pg5>.

I. V. Anikina, A. Bogoliubov Laboratory of Theoretical Physics, JINR, Dubna, Russia, Institute of Modern Physics, Chinese Academy of Sciences, Lanzhou, P.R. China, Southern Center for Nuclear Science Theory, Institute of Modern Physics, Chinese Academy of Sciences

Bogoliubov Laboratory of Theoretical Physics · Institute of Modern Physics Chinese Academy of Sciences Lanzhou · Southern Center for Nuclear Science Theory Institute of Modern Physics Chinese Academy of Sciences Huizhou

hep-th, math-ph, math.FA, math.MP

Submitted: 2026-11-05

Updated: 2026-11-05

Comments: 9 pages in JHEP style

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 51/100

The gist: In this paper we use principles of non-symmetric integral geometry to extend relations traced from the Fourier slice theorem and dual Radon transform for an alternative reconstruction procedure that…

Key concepts

Dual Radon Transform
This transform relates a function G to its integral over lines with an angular dependence. It involves integrating over both the line parameters (like position and angle) and the function's parameters, offering a different perspective on reconstructing images from line integrals.
Generalized Inverse Fourier Slice Theorem
The standard theorem for inverting Fourier slices is modified here. Because the integration measure has symmetry breaking, this generalized version shows that the relationship between the Fourier domain ($ ilde{F}[f]$) and the Radon domain ($R[f]$) is more intricate than usual.
Non-symmetric Integral Geometry
This approach uses principles from integral geometry that do not rely on full symmetry. It acknowledges that when an original function lacks symmetry, the angular integration measure is restricted, which complicates the standard universal inversion process and requires a more nuanced mathematical framework.

Terminology

Summary

In this paper we use principles of non-symmetric integral geometry to extend relations traced from the Fourier slice theorem and dual Radon transform for an alternative reconstruction procedure that impacts image reconstruction processes. p. 2

Where the paper says it

In the presented paper, we use the non-symmetric, extended integral geometry to improve the relations emanated from the Fourier slice theorem and the dual Radon transform. The proposed generalization impacts on the image reconstruction process [10, 11].

The derived relations confirm the complexity of the inversion operator due to the symmetry breaking of integration measure. p. 0
Where the paper says it

The latter impacts entirely on the image reconstruction process. The derived relations confirm the complexity of the inversion operator thanks for the symmetry breaking of integration measure.

Introduction and Motivation

Integral geometry is backed by symmetry properties of measures on geometrical spaces, and the direct Radon transform is a main object in integral geometry. The inverse Radon transforms are used to reconstruct images, but they often face ill-posedness necessitating regularization. p. 1

Where the paper says it

Indeed, the inverse Radon transforms (IRT) are being used to reconstruct the image of an object under investigation. As a rule, the Radon inversion meets the problems dictated by the certain ill-posedness.

The old-fashion Radon inversion realized in odd and even dimensions separately depends on the parity of space dimension because angular integration is always performed over the full region (0, 2π). p. 1
Where the paper says it

In contrast to [5–8], the old-fashion Radon inversion has been realized in the odd and even dimension of space separately and independently each other. The dependence of old-fashion inversion methods on the parity of space dimension can be traced from the fact that the angular – 1 –

If the original homogeneous outset function is well-localized and lacks symmetry properties, the Radon ϕ-angular dependence receives restrictions instead of the full region (0, 2π). p. 2
Where the paper says it

in the full interval (0, 2π) (see Courant-Hilbert’s identities in [9]). As demonstrated, if the original homogeneous outset function, which is under DRT, is welllocalized in the space and does not possess the symmetry properties, the Radon ϕ-angular dependence receives the definite restrictions, instead of the full region (0, 2π).

This angular restriction makes the universal Radon inversion a complex transformation involving contributions fS and fA induced by the complex integration measure. p. 2
Where the paper says it

Due to these angular restrictions, the universal Radon inversion becomes a complex transformation. It involves two contributions fS and fA, where fA (this is the additional term compared to the standard inversion) is induced by the complex integration measure.

The presented principles of Radon inversion extend and generalize integral geometry, which is called non-integral geometry. p. 2
Where the paper says it

The presented principles of Radon inversion extend and generalize essentially the integral geometry method. It is worth to notice that the new principles of the extended integral geometry method are previously called as non-integral geometry [7, 8].

Dual Radon Transforms and Fourier Space

The dual Radon transform is defined as R∗G = Zωdµ(τ,ϕ)G(τ,ϕ)δ τ − ⟨⃗nϕ,⃗x⟩. After τ-integration, we obtain R∗G = Zωϕ(dϕ)G(⟨⃗nϕ,⃗x⟩,ϕ). Assuming G is the Radon image R(τ,ϕ), we write the Fourier image of the dual R-transformation as FR∗R[f] = Z+∞−∞(d2⃗x)e−i⟨⃗q,⃗x⟩R∗Rf. The inverse Fourier transform is R∗Rf = Z+∞−∞(d2⃗q)ei⟨⃗q,⃗x⟩FR∗R[f]. In the polar system, the Fourier radial λ-component of Fourier variables varies in the standard interval (0, +∞). p. 3

Where the paper says it

We begin with the inverse Fourier transform given by f(⃗x) = Z +∞ −∞ (d n ⃗q)e i⟨⃗q,⃗x⟩F[ f ](⃗q). 5) In the polar system the radial λ-component of Fourier variables, ⃗q = λ⃗nϕ, varies in the standard interval (0, +∞).

The analytical extension of λ up to the full interval depends on the symmetry property of the integration ϕ-measure. p. 3
Where the paper says it

That is, in the polar system we have the following: f(⃗x) = Z +∞ 0 dλ λ Z ωϕ (dϕ)e iλ⟨⃗nϕ ,⃗x⟩F[ f ](λ,ϕ). 6) The analytical extension of λ up to the full interval (−∞, +∞) depends on the symmetry property of the integration ϕ-measure.

Generalized Inverse Fourier Slice Theorem

The inverse Fourier slice theorem states Ff = Z+∞−∞(dτ)e−iλ τRf. Inverting this transformation leads to a generalized inverse Fourier slice theorem: Z+∞0(dλ)eiλτ˜Ff = Z+∞−∞(dτ)iτ˜−τ +iεRf. This representation is more complicated than the standard inversion because the Fourier radial λ-component and the Radon radial τ-component vary in different regions. p. 4

Where the paper says it

Since in (3. 1) the Fourier radial λ-component and the Radon radial τ-component vary in the different regions (i.

Alternative Reconstruction Procedure

The Fourier transform of R∗R[f] is given by FR∗R[f] = 1/⃗q Ff−iπ R[f]⊛Q(⃗q). This generalized representation is inserted into the standard inverse Fourier transform to get the new and extended representation for alternative inversion procedure: f(⃗x) = Z+∞−∞(d2⃗q)ei⟨⃗q,⃗x⟩/vec q FR∗R[f] + iπ R[f]⊛Q(⃗q). The complexity is also extracted from the Q-function, which is given by Q(η,ϕ;⃗q) = Zωφ(dφ)Z+∞0(dr)r e−iλr cos(ϕ−φ)P r cos(ϕ −φ −η). p. 5

Where the paper says it

In addition, the complexity can be also extracted from Q-function. Indeed, we have the following Q(η,ϕ;⃗q) ⃗q=λ⃗nϕ = Z ωφ (dφ) Z +∞ 0 (dr)r e−iλr cos(ϕ−φ) P r cos(ϕ −φ)−η , (5.

This Q-function shows that for the polar system, Q(η,ϕ;⃗q=λ⃗nϕ) ∼ Z+∞0(dr)e−iCr r/r-B ⇒ Z+∞0(dr)e−iCr = δ−(C) ∈ C. p. 5
Where the paper says it

2) where the polar system for the position space, ⃗x = r⃗nφ = (r, φ), has been used. Hence, we can readily see that Q(η,ϕ;⃗q) ⃗q=λ⃗nϕ ∼ Z +∞ 0 (dr)e −iCr r r −B ⇒ Z +∞ 0 (dr)e −iCr = δ−(C) ∈ C.

Conclusion

The study demonstrates that the ϕ-angular restrictions lead to improvements in the inverse Fourier slice theorem compared to the standard inversion. The generalized representation for the Fourier transform of the dual Radon image is important for obtaining an alternative inversion procedure. p. 5

Where the paper says it

Therefore, we derive the generalized representation for the Fourier transform of the dual Radon image, see (4. This generalized representation plays an important role for the derivation of an alternative inversion procedure used in the image reconstruction process.

Finally, a new and extended representation for the alternative inversion procedure needed for image reconstruction has been derived. p. 5
Where the paper says it

Hence, we can readily see that Q(η,ϕ;⃗q) ⃗q=λ⃗nϕ ∼ Z +∞ 0 (dr)e −iCr r r −B ⇒ Z +∞ 0 (dr)e −iCr = δ−(C) ∈ C. 3) We thus derive the new and extended representation for the alternative inversion procedure needed for the image reconstruction.

Appendix on Well-localized Functions

For the well-localized one-dimensional function f(x) = e−x2 with x ∈ R1, its Fourier image is Ff = Z+∞−∞(dx)e−(x+iq/2)2−q2/4. If the Fourier variable q is small, the well-localized Fourier image reads Ff = Ne−q2/4. However, if q is large enough, the case corresponds to Ff = Z+∞−∞(dx)e−x2cos(qx), and lim q→∞ Ff ∼ lim M→∞ cos(M) ≠ 0. The characteristic function in R1 f(x) = Θx ≤ a has the Fourier image Ff = Z+a−a(dx)cos(qx), and lim M→∞ sin(M) ≠ 0. This fact is important when dealing with the inclusion of noise effect. p. 7

Where the paper says it

6) decreases fast but it is not equal to zero. p. 7

Where the paper says it

6) One can see that (A. 6) decreases fast but it is not equal to zero.

This fact becomes to be very important while we deal with the inclusion of the noise effect.

References

[1] I. M. Gelfand and G. p. 7

Where the paper says it

References [1] I. Gelfand and G.

E. Shilov, “Generalized Functions Vol 1 Properties And Operations,” Academic Press, 1964, ISBN-0-12-279501-6

[2] I.M. Gelfand, M. p. 7

Where the paper says it

References [1] I. Gelfand and G.

I. Graev, N.Ya. Vilenkin, “Generalized Functions, Volume 5: Integral Geometry and Representation Theory,” AMS Chelsea Publishing: An Imprint of the American Mathematical Society, 449 pp (1966)

[3] I.M. Gelfand, S. p. 7

Where the paper says it

References [1] I. Gelfand and G.

G. Gindikin, M. p. 7
Where the paper says it

Gelfand, S. Gindikin, M.

I. Graev, “Selected topics in integral geometry,” Volume 220, Translations of Mathematical Monographs, American Mathematical Society, 2003

[4] S.R. Deans, “The Radon Transform and Some of Its Applications,” Wiley 299 p (1983)

[5] I. V. Anikin and L. p. 0

Where the paper says it

China cSouthern Center for Nuclear Science Theory, Institute of Modern Physics, Chinese Academy of Sciences, Huizhou, P. China E-mail: anikin@theor.

Szymanowski, Phys. p. 7
Where the paper says it

Anikin and L. Szymanowski, Phys.

Rev. D 100, no.9, 094034 (2019) doi:10. p. 7
Where the paper says it

Szymanowski, Phys. 9, 094034 (2019) doi:10.

1103/PhysRevD. p. 7
Where the paper says it

9, 094034 (2019) doi:10. 1103/PhysRevD.

100.094034 [arXiv:1909. p. 7
Where the paper says it

1103/PhysRevD. 094034 [arXiv:1909.

06876 [hep-ph]]

[6] I. V. Anikin and X. p. 0

Where the paper says it

China cSouthern Center for Nuclear Science Theory, Institute of Modern Physics, Chinese Academy of Sciences, Huizhou, P. China E-mail: anikin@theor.

Chen, Mod. Phys Lett A 39, no 38, 2450181 (2024) doi:10. p. 8
Where the paper says it

Anikin and X. 38, 2450181 (2024) doi:10.

1142/S0217732324501815 [arXiv:2405. p. 8
Where the paper says it

38, 2450181 (2024) doi:10. 1142/S0217732324501815 [arXiv:2405.

14897 [physics. p. 8
Where the paper says it

1142/S0217732324501815 [arXiv:2405. 14897 [physics.

comp-ph]]

[7] I. V. Anikin, Natural Sci Rev 2, no 5, 100501 (2025) doi:10. p. 8

Where the paper says it

Anikin, Natural Sci. 5, 100501 (2025) doi:10.

54546/NaturalSciRev. p. 8
Where the paper says it

5, 100501 (2025) doi:10. 54546/NaturalSciRev.

100501 [arXiv:2506. p. 8
Where the paper says it

54546/NaturalSciRev. 100501 [arXiv:2506.

18911 [math. p. 8
Where the paper says it

100501 [arXiv:2506. 18911 [math.

FA]]

[8] I. V. Anikin, Mod Phys Lett A 40, no 32, 2550141 (2025) doi:10. p. 8

Where the paper says it

Anikin, Mod. 32, 2550141 (2025) doi:10.

1142/S021773232550141X [arXiv:2504. p. 8
Where the paper says it

32, 2550141 (2025) doi:10. 1142/S021773232550141X [arXiv:2504.

01744 [math. p. 8
Where the paper says it

1142/S021773232550141X [arXiv:2504. 01744 [math.

CA]]

[9] R. Courant and D. p. 1

Where the paper says it

The universality of inverse Radon transform means the independence of inversion on the space dimension. Notice that the obtained universal representation for the Radon inversion has been derived with the help of the appropriated regularization procedure, used in the generalized function (distribution) theory [1–3], and, what is important, without referring Courant-Hilbert’s identities.

Hilbert, “Methods of Mathematical Physics,” Vo II, Interscience, New York (1962)

[10] I. V. Anikin, A.I. Anikina and O. p. 0

Where the paper says it

PREPRINT TYPESET IN JHEP STYLE Non-symmetric integral geometry: alternative reconstruction procedure with dual Radon transform I. Anikina,b,c aBogoliubov Laboratory of Theoretical Physics, JINR, Dubna, Russia b Institute of Modern Physics, Chinese Academy of Sciences, Lanzhou, P.

I. Streltsova “Reconstruction Problem: Computational Scheme,” in preparation

[11] M.Ya Marusina, AO Kazacheeva “Modern methods of tomography lecture textbook," ITMO University, St Petersburg, (2006)

--- Page 8 ---

(a) a new integration variable z is a real one due to q < M < ∞. That is, q is rather a small (or finite) Fourier variable and it can be neglected. p. 7

Where the paper says it

• (a) a new integration variable z is a real one due to q < M < ∞. That is, q is rather a small (or finite) Fourier variable and it can be neglected.

Moreover, this z varies into the interval (−∞, ∞);

--- Page 8 ---

(b) a new integration variable z becomes a complex one and varies also in the infinite interval provided the Fourier variable q ∼ [∞], i.e. q ∼ M → ∞ (q is a large enough). p. 7

Where the paper says it

Moreover, this z varies into the interval (−∞, ∞); • (b) a new integration variable z becomes a complex one and varies also in the infinite interval provided the Fourier variable q ∼ [∞], i. q ∼ M → ∞ (q is a large enough).

--- Page 9 ---

By direct calculations of (A.2), one can readily see that the case (a) leads to the well-localized Fourier image: Ff = Ne−q2/4.

--- Page 9 ---

While the case (b) corresponds to the Fourier image of the well-localized function that is not localized at all. Indeed, we have the following: Ff = Z+∞−∞(dx)e−x2cos(qx).

--- Page 9 ---

One can see that (A.6) decreases fast but it is not equal to zero.

--- Page 8 ---

The other similar example is given by the characteristic function in R1: f(x) = Θx ≤ a.

--- Page 8 ---

Having calculated, in this case the Fourier image reads Ff = Z+a−a(dx)cos(qx).

--- Page 8 ---

This fact becomes to be very important while we deal with the inclusion of the noise effect.

--- Page 9 ---

The origin function e−x2 can lead to the localized Fourier image owing to the restricted (or finite) Fourier variable q.

--- Page 9 ---

The other similar example is given by the characteristic function in R1: f(x) = Θx ≤ a.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Alternative Image Reconstruction Procedure: The system can perform image reconstruction using equation (5.1), which is given by f(x) = Z +infinity-infinity (dq)e i n FR∗R[f] + i pi R[f]⊛Q(q). This new representation manifests complexity through the Q-function, which relates to position space variables.

  2. Improved Inversion of Dual Radon Transform: The system can utilize the generalized inverse Fourier slice theorem, equation (3.2), to invert the transformation, which is presented as Z +infinity 0 (dλ)e iλ(τ˜−τ) = Z +infinity-infinity (dτ) i(˜−τ − τ + i epsilon) Rf.

  3. Handling Non-Symmetric Integration Measures: The system can account for the symmetry breaking effects in the integration measure by employing non-integral geometry principles, which is described as complete and incomplete integration measures. This allows the method to be more suitable for original objects that are naturally non-symmetric ones.

Abstract

In the note, we use the principles of non-symmetric integral geometry (non-integral geometry), related to the non-symmetric restricted integration measure, to extend the well-known relations traced from the Fourier slice theorem and the dual Radon transform. The proposed generalization is a very useful for the alternative restoration of the outset function. The latter impacts entirely on the image reconstruction process. The derived relations confirm the complexity of the inversion operator thanks for the symmetry breaking of integration measure.

Sources

Related papers