Topological derivative for a fast identification of short, linear perfectly conducting cracks with inaccurate background information
Won-Kwang Park · 2025
Read Topological derivative for a fast identification of short, linear perfectly conducting cracks with inaccurate background information by Won-Kwang Park (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-2504-19485.txt.
Introduction
Won-Kwang Park develops a fast imaging method to find short, straight, perfectly conducting cracks hidden inside a material. The method rests on the topological derivative, which gauges where a tiny new flaw would most change the measured field. A twist here is that the background material properties are not known exactly, since they shift with frequency and heat.
Park proposes a normalized imaging function that can still show a crack exists under wrong background values. The crack's apparent position shifts, though, in a way he explains with a formula involving a Bessel function, the crack length, and the wrong background wavenumber. Tests on noisy synthetic data confirm the theory.
Catalog record
| CREATOR | Won-Kwang Park |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-19485
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Topological derivative for a fast identification of short, linear perfectly conducting cracks with inaccurate background information Won-Kwang Park Abstract In this study, we consider a topological derivative-based imaging technique for the fast identification of short, linear perfectly conducting cracks completely embedded in a two-dimensional homogeneous domain with smooth boundary. Unlike conventional approaches, we assume that the background permittivity and permeability are unknown due to their dependence on frequency and temperature, and we propose a normalized imaging function to localize cracks. Despite inaccuracies in background parameters, application of the proposed imaging function enables to recognize the existence of crack but it is still impossible to identify accurate crack locations. Furthermore, the shift in crack localization of imaging results is significantly influenced by the applied background parameters. In order to theoretically explain this phenomenon, we show that the imaging function can be expressed in terms of the zero-order Bessel function of the first kind, the crack lengths, and the applied inaccurate background wavenumber corresponding to the applied inaccurate background permittivity and permeability. Various numerical simulations results with synthetic data polluted by random noise validate the theoretical results. 1 Introduction Accurate and efficient identification of unknown defects from scattering phenomena by applying microwaves is of paramount importance in various engineering and physical applications, including non-destructive testing [ 25 , 42 , 47 ] , structural health monitoring [ 22 , 26 , 29 ] , biomedical imaging [ 2 , 3 , 19 ] , and radar technology [ 9 , 21 , 46 ] . Due to the fundamental nonlinearity and ill-posedness of the problem, various iterative (quantitative) inversion techniques have been developed to address microwave imaging challenges. Examples include Newton-type shape reconstruction for arc-like cracks [ 31 ] , Gauss–Newton method for breast cancer imaging [ 44 ] , Born iterative approach for brain stroke detection [ 28 ] , level set method for retrieving shape of unknown scatterers [ 23 ] , Newton–Kantorovich method for arm and thorax imaging [ 33 ] , and optimal control approach for shape reconstruction of extended scatterers [ 4 ] . Although these iterative-based inversion techniques have been successfully applied to various real-world microwave imaging problems, several critical issues such as non-convergence phenomena, entrapment in local minima, and significant computational costs due to the large number of iterations, may arise if the iterative process is initiated without a good initial guess and prior information of unknown targets. Moreover, because iterative-based techniques are highly sensitive to measurement noise, this can lead to substantial errors in the reconstructed results so that appropriate regularization strategies must be adopted. Hence, for obtaining a good initial guess or prior information (location, shape, etc.) of unknown objects, various non-iterative (qualitative) inversion techniques have been investigated. Among the many available non-iterative techniques, topological derivative-based imaging methods have drawn significant attention. Originally, this was originally investigated for the shape optimization problem [ 1 , 16 , 18 , 24 , 36 , 45 ] , but its application for fast identification of unknown objects in inverse scattering problem has been demonstrated successfully; for example, identification of small defects [ 11 , 12 ] , fast imaging of thin electromagnetic inhomogeneities [ 37 , 38 ] , shape imaging of three dimensional obstacles [ 32 ] , flaw detection in welding joints [ 35 ] , object localization from Fresnel experimental dataset [ 15 ] , and infrared thermography for defect identification inside thin plates [ 41 ] . We also refer to [ 5 , 13 , 14 , 17 , 27 ] for various remarkable results about the topological derivatives. In general, the effectiveness of topological derivative methods hinges on the accurate knowledge of background material properties—specifically, the permittivity, permeability, and conductivity of the medium. However, in real-world applications, such parameters are often inaccurately known due to their sensitivity to environmental variations such as temperature and operating frequency. Related works can be found in [ 10 , 20 , 30 , 34 , 43 ] and we refer to Figure 1 . This leads to critical issues in practical implementations, where even slight deviations from true background values can inaccurate imaging result. However, to the best of our knowledge, a reliable mathematical theory has not been investigated to explain this phenomenon. 0.5\text{\,}\mathrm{GHz} 1\text{\,}\mathrm{GHz} 1.5\text{\,}\mathrm{GHz} 60 61 62 63 blood 0.5\text{\,}\mathrm{GHz} 1\text{\,}\mathrm{GHz} 1.5\text{\,}\mathrm{GHz} 11.2 11.4 fat 0.5\text{\,}\mathrm{GHz} 1\text{\,}\mathrm{GHz} 1.5\text{\,}\mathrm{GHz} 84.2 84.4 84.6 water Figure 1: Permittivity values of blood (left), fat (center), and water (right) between f=$0.5\text{\,}\mathrm{GHz}$ and f=$1.5\text{\,}\mathrm{GHz}$ . To address this issue, we consider the application of topological derivative for identifying a set of two-dimensional short, linear perfectly conducting cracks completely embedded in a homogeneous medium, even when accurate background permittivity or permeability values are unknown. To this end, we introduce a topological derivative based imaging function based on a misfit functional with inaccurate permittivity and permeability values. In order to explain some phenomena including the appearance of inaccurate location of cracks, we carefully investigate the structure of the imaging function by establishing its relationship with crack length, both accurate and inaccurate wavenumbers, and the zero-order Bessel function. This is based on the fact that the boundary measurement data can be expressed by the asymptotic expansion formula in the presence of small cracks. To validate the theoretical results, numerical simulation results with boundary measurement data corrupted by random noise in the existence of single and multiple short, linear cracks are exhibited. This paper is organized as follows: In Section 2 , we introduce the basic concept of the forward problem in the presence of a set of short, linear perfectly conducting cracks and design topological derivative based imaging function with inaccurate background wavenumber corresponding to the applied permittivity and permeability values. In Section 3 , we explore mathematical structure of the designed imaging function and discuss its various properties including the appearance of inaccurate results. In Section 4 , we present numerical simulation results to validate the theoretical result. In Section 5 , a short conclusion including future directions is provided. 2 Topological derivative based imaging Let \Gamma_{m} be a linear perfectly conducting crack with length 2\ell_{m} and center \mathbf{x}_{m} and \Gamma denotes the collection of \Gamma_{m} . Throughout this paper, all \Gamma_{m} are completely embedded in a homogeneous material \Omega\subset\mathbb{R}^{2} with smooth boundary \partial\Omega\in\mathcal{C}^{2} , permittivity \varepsilon_{\mathrm{b}} , and permeability \mu_{\mathrm{b}} at given angular frequency \omega=2\pi f . Throughout this paper, we assume that \Omega is a subset of anechoic chamber; thus, \mu_{\mathrm{b}}=4\pi\times${10}^{-7}\text{\,}\mathrm{H}$ , \varepsilon_{\mathrm{b}}=$8.854\text{\times}{10}^{-12}\text{\,}\mathrm{F}$ , and the background conductivity \sigma_{\mathrm{b}}\approx 0 . By denoting k=\omega\sqrt{\varepsilon_{\mathrm{b}}\mu_{\mathrm{b}}} and \lambda as the background wavenumber and given positive wavelength, respectively, we assume that k|\mathbf{x}_{m}-\mathbf{x}_{m^{\prime}}|\gg\frac{1}{4}\quad\text{and}\quad\ell_{m}<\frac{\lambda}{2}, (1) for m,m^{\prime}=1,2,\ldots,M and m\neq m^{\prime} . These conditions means that \Gamma_{m} are sufficiently separated from each other and the their lengths are short. Furthermore, we assume that \Gamma_{m} does not touch the \partial\Omega i.e., there is a positive constant C such that \dist(\Gamma_{m},\partial\Omega)\geq C. In the presence of \Gamma , let u^{(n)}(\mathbf{x};k) be the time-harmonic total field satisfying the boundary value problem \left\{\begin{array}[]{rcl}\vskip 6.0pt plus 2.0pt minus 2.0pt\triangle u^{(n)}(\mathbf{x};k)+k^{2}u^{(n)}(\mathbf{x};k)=0&\mbox{in}&\Omega\backslash\overline{\Gamma}\\ u^{(n)}(\mathbf{x};k)=0&\mbox{on}&\Gamma\\ \vskip 6.0pt plus 2.0pt minus 2.0pt\cr\displaystyle\frac{\partial u^{(n)}(\mathbf{x};k)}{\partial\boldsymbol{\nu}(\mathbf{x})}=\frac{\partial\exp(ik\boldsymbol{\theta}_{n}\cdot\mathbf{x})}{\partial\boldsymbol{\nu}(\mathbf{x})}=g^{(n)}(\mathbf{x})\in L^{2}(\partial\Omega)&\mbox{on}&\partial\Omega,\\ \end{array}\right. (2) where \boldsymbol{\nu}(\mathbf{x}) is the outward normal to \mathbf{x}\in\partial\Omega and \boldsymbol{\theta}_{n}\in\mathbb{S}^{1} , n=1,2,\ldots,N , denotes the propagation direction.