Inverse problems, Ill-posedness and regularization – an illustrative example

SummaryWhenever one is confronted with the necessity to measure some quantities, which are not accessible directly, however, are linked via a mathematical model to some measurement data, one has to solve an inverse problem. In this context we speak of a direct problem, when expected measurement data are calculated from a mathematical model, when the not directly accessible quantities are given and, on the other hand, of an inverse problem, when these quantities are calculated from measured data via the mathematical model. In this paper the principles of inverse problems are explained on the basis of a one-dimensional image restoration problem, which is a linear problem and hence is easy to understand. Furthermore, the term ill-posedness will be explained and some possibilities to attain reasonable solutions to ill-posed problems are discussed.ZusammenfassungWann immer man mit der Aufgabe konfrontiert ist, nicht-direkt erfassbare Daten, die aber über ein mathematisches Modell mit direkt messbaren Größen verlinkt sind, messen zu müssen, muss man ein inverses Problem lösen. In diesem Zusammenhang spricht man von einem direkten Problem, wenn man mit Vorgabe der nicht-direkt erfassbaren Größen über ein mathematisches Modell die zu erwartenden Messdaten berechnet, und umgekehrt von einem inversen Problem, wenn man aus den Messdaten – wieder über den Umweg über dasselbe mathematische Modell – die nicht-direkt erfassbaren Größen berechnet. In diesem Beitrag werden die Grundzüge von inversen Problemen anhand eines linearen eindimensionalen Image-Restoration-Problems dargelegt. Weiters werden der Ausdruck Schlechtgestelltheit erklärt und Möglichkeiten, zu annehmbaren Lösungen für schlechtgestellte Probleme zu kommen, diskutiert.

[1]  Daniel Watzenig,et al.  Capacitive sensor signal processing based on decentralized Kalman filtering , 2003, IEEE International Conference on Industrial Technology, 2003.

[2]  Michel Barlaud,et al.  Deterministic edge-preserving regularization in computed imaging , 1997, IEEE Trans. Image Process..

[3]  Bernhard Brandstätter,et al.  A FEM‐BEM approach using level‐sets in electrical capacitance tomography , 2005 .

[4]  G. Holler,et al.  Adaptive regularization parameter adjustment for reconstruction problems , 2004, IEEE Transactions on Magnetics.

[5]  Daniel Watzenig,et al.  Reconstruction of inhomogeneities in fluids by means of capacitance tomography , 2003 .

[6]  C. Vogel Computational Methods for Inverse Problems , 1987 .

[7]  J. Hadamard,et al.  Lectures on Cauchy's Problem in Linear Partial Differential Equations , 1924 .

[8]  Daniel Watzenig,et al.  Accurate estimation of angular position and speed for a capacitive sensor based on an extended Kalman filter approach , 2003, 1st International Workshop on Robotic Sensing, 2003. ROSE' 03..

[9]  Robert Tibshirani,et al.  An Introduction to the Bootstrap , 1994 .

[10]  William R B Lionheart,et al.  Reconstruction Algorithms for Permittivity and Conductivity Imaging , 2001 .

[11]  Stanley Osher,et al.  Total variation based image restoration with free local constraints , 1994, Proceedings of 1st International Conference on Image Processing.

[12]  Curtis R. Vogel,et al.  Iterative Methods for Total Variation Denoising , 1996, SIAM J. Sci. Comput..

[13]  Eung Je Woo,et al.  Fast Static Image Reconstruction using Adaptive Mesh Grouping Method in EIT , 1997, Proceedings of the 19th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 'Magnificent Milestones and Emerging Opportunities in Medical Engineering' (Cat. No.97CH36136).

[14]  Wuqiang Yang,et al.  Prior-online iteration for image reconstruction with electrical capacitance tomography , 2004 .

[15]  Todd Pugsley,et al.  The influence of permittivity models on phantom images obtained from electrical capacitance tomography , 2002 .

[16]  D. Watzenig,et al.  Offset compensation for capacitive angular position sensors by evaluating the Kalman filter innovation sequence , 2004, 2004 IEEE International Conference on Industrial Technology, 2004. IEEE ICIT '04..

[17]  Wolfgang Osten,et al.  Introduction to Inverse Problems in Imaging , 1999 .

[18]  Willis J. Tompkins,et al.  Comparing Reconstruction Algorithms for Electrical Impedance Tomography , 1987, IEEE Transactions on Biomedical Engineering.

[19]  B. Brandstatter,et al.  Jacobian calculation for electrical impedance tomography based on the reciprocity principle , 2003 .

[20]  J. Sethian,et al.  Fronts propagating with curvature-dependent speed: algorithms based on Hamilton-Jacobi formulations , 1988 .

[21]  Supachai Bangliang,et al.  The use of simultaneous iterative reconstruction technique for electrical capacitance tomography , 2000 .

[22]  Brian S. Hoyle,et al.  Electrical capacitance tomography for flow imaging: system model for development of image reconstruction algorithms and design of primary sensors , 1992 .

[23]  K. Y. Kim,et al.  Dynamic Electrical Impedance Tomography with Known Internal Structures , 2003 .

[24]  G. Holler,et al.  Comparison of measurement results and simulations based on finite element method for an electrical capacitance tomography system , 2004, ISA/IEEE Sensors for Industry Conference, 2004. Proceedings the.

[25]  S. Osher,et al.  Algorithms Based on Hamilton-Jacobi Formulations , 1988 .

[26]  Xiaodong Jia,et al.  Generic reality visualization modelling from process tomography sensor data in process design , 2004 .

[27]  P.A. Karjalainen,et al.  A Kalman filter approach to track fast impedance changes in electrical impedance tomography , 1998, IEEE Transactions on Biomedical Engineering.

[28]  Mario Bertero,et al.  Introduction to Inverse Problems in Imaging , 1998 .

[29]  J. Hadamard,et al.  Lectures on Cauchy's Problem in Linear Partial Differential Equations , 1924 .