×

Approximation by \(\Gamma\)-convergence of a curvature-depending functional in visual reconstruction. (English) Zbl 1098.49012

Let \(\Omega\) be a bounded open subset of \({\mathbb R}^2\), and let \(g\in L^\infty(\Omega)\). The paper proposes an approximation, in the sense of \(\Gamma\)-convergence, of the functional \[ {\mathcal G}(u,C,P)=\#(P)+\int_C(1+\kappa^2)\,d{\mathcal H}^1+\int_{\Omega\backslash(C\cup P)}| \nabla u| ^2\,dx+\int_\Omega| u-g| ^2\,dx, \] where \(C\) is a family of curves, \(P\) is the set of the endpoints of the curves of \(P\), \(\#(P)\) is the number of points in \(P\), \(\kappa\) is the curvature, and \({\mathcal H}^1\) is the one-dimensional Hausdorff measure.
The approximating functionals are of “elliptic type”, so, at least in principle, numerically more tractable. They are of the form \[ \begin{split} {\mathcal G}_\varepsilon(u,s,w)={1\over4\pi b_0}\int_\Omega\left({1\over\varepsilon}+\varepsilon\left(\text{div}{\nabla w\over| \nabla w| }\right)^2\right)\left(\zeta_\varepsilon| \nabla w| ^2+{w^2(1-w)^2\over\zeta_\varepsilon}\right)\,dx+\\ +{1\over2b_0}\int_\Omega w^2\left(1+\left(\text{div}{\nabla s\over| \nabla s| }\right)^2\right)\left(\zeta_\varepsilon| \nabla s| ^2+{s^2(1-s)^2\over\zeta_\varepsilon}\right)\,dx+\int_\Omega s^2| \nabla u| ^2\,dx+\\+\int_\Omega| u-g| ^2dx+{1\over\mu_\varepsilon}\int_\Omega((1-s)^2+(1-w)^2)dx,\end{split} \] for suitable \(\zeta_\varepsilon\) and \(\mu_\varepsilon\) tending to 0, and \(b_0>0\).
Connections with a related conjecture by E. De Giorgi are also discussed.

MSC:

49J45 Methods involving semicontinuity and convergence; relaxation
49Q20 Variational problems in a geometric measure-theoretic setting
49Q15 Geometric measure and integration theory, integral and normal currents in optimization
68U10 Computing methodologies for image processing
Full Text: DOI

References:

[1] Alberti, Indiana Univ Math J
[2] Amar, Ann Sc Norm Super Pisa Cl Sci (5) 2 pp 151– (2003)
[3] Ambrosio, Boll Un Mat Ital B (7) 3 pp 857– (1989)
[4] ; ; Functions of bounded variation and free discontinuity problems. Oxford Mathematical Monographs. Clarendon Press, New York, 2000.
[5] Ambrosio, Comm Pure Appl Math 43 pp 999– (1990)
[6] Functions of bounded variation over non-smooth manifolds and generalized curvatures. Variational methods for discontinuous structures (Como, 1994), 135-142. Progress in Nonlinear Differential Equations and Their Applications 25. Birkhäuser, Basel–Boston, 1996. · doi:10.1007/978-3-0348-9244-5_13
[7] ; Mathematical problems in image processing. Applied Mathematical Sciences, 147. Springer, New York–Berlin, 2002.
[8] Bellettini, J Convex Anal 4 pp 91– (1997)
[9] Bellettini, Ann Scuola Norm Sup Pisa Cl Sci (4) 20 pp 247– (1993)
[10] Bellettini, Ann Inst H Poincaré Anal Non Linéaire 21 pp 839– (2004)
[11] ; ; Ginzburg-Landau vortices. Progress in Nonlinear Differential Equations and Their Applications, 13. Birkhäuser, Basel–Boston, 1994. · doi:10.1007/978-1-4612-0287-5
[12] ; Visual reconstruction. MIT Press Series in Artificial Intelligence. MIT Press, Cambridge, Mass., 1987.
[13] Bourdin, Numer Math 85 pp 609– (2000)
[14] Approximation of free-discontinuity problems. Lectures Notes in Mathematics, 1694. Springer, Berlin, 1998. · Zbl 0909.49001 · doi:10.1007/BFb0097344
[15] {\(\Gamma\)}-convergence for beginners. Oxford Lecture Series in Mathematics and Its Applications, 22. Oxford University Press, Oxford, 2002. · doi:10.1093/acprof:oso/9780198507840.001.0001
[16] Braides, Interfaces Free Bound 4 pp 345– (2002)
[17] Chambolle, Comm Partial Differential Equations 22 pp 811– (1997)
[18] Coscia, Nonlinear Anal 39 pp 711– (2000)
[19] An introduction to {\(\Gamma\)}-convergence. Progress in Nonlinear Differential Equations and Their Applications, 8. Birkhäuser, Basel–Boston, 1993. · doi:10.1007/978-1-4612-0327-8
[20] Dal Maso, Acta Math 168 pp 89– (1992)
[21] Some remarks on {\(\Gamma\)}-convergence and least squares methods. Composite media and homogenization theory (Trieste, 1990), 135-142. Progress in Nonlinear Differential Equations and Their Applications, 5. Birkhäuser, Boston, 1991. · doi:10.1007/978-1-4684-6787-1_8
[22] De Giorgi, Arch Rational Mech Anal 108 pp 195– (1989)
[23] Esedoglu, J Math Imaging Vision 18 pp 7– (2003)
[24] Esedoglu, European J Appl Math 13 pp 353– (2002)
[25] Variational problems with concentration. Progress in Nonlinear Differential Equations and Their Applications, 36. Birkhäuser, Basel, 1999. · doi:10.1007/978-3-0348-8687-1
[26] Friesecke, Comm Pure Appl Math 55 pp 1461– (2002)
[27] Mantegazza, J Differential Geom 43 pp 807– (1996)
[28] March, Image and Vision Computing 10 pp 30– (1992)
[29] March, Image and Vision Computing 15 pp 705– (1997)
[30] Modica, Boll Un Mat Ital B (5) 14 pp 285– (1977)
[31] ; Variational methods in image segmentation. Progress in Nonlinear Differential Equations and Their Applications, 14. Birkhäuser, Basel–Boston, 1995. · doi:10.1007/978-1-4684-0567-5
[32] Mumford, Comm Pure Appl Math 42 pp 577– (1989)
[33] Nesi, Image and Vision Computing 11 pp 419– (1993)
[34] ; ; Filtering, segmentation and depth. Lecture Notes in Computer Science, 662. Springer, Berlin, 1993. · Zbl 0801.68171 · doi:10.1007/3-540-56484-5
[35] ed. Geometry driven diffusion in computer vision. Kluwer, Dordrecht, 1994. · Zbl 0832.68111 · doi:10.1007/978-94-017-1699-4
[36] Sandier, Comm Pure Appl Math 57 pp 1627– (2004)
[37] Uses of elliptic approximations in computer vision. Variational methods for discontinuous structures (Como, 1994), 19-34, Progress in Nonlinear Differential Equations and Their Applications, 25. Birkhäuser, Basel–Boston, 1996. · doi:10.1007/978-3-0348-9244-5_3
[38] Shah, J Visual Commun Image Repres 13 pp 36– (2002)
[39] Shah, IEEE Trans Image Process 5 pp 1243– (1996)
[40] Sternberg, Arch Rational Mech Anal 101 pp 209– (1988)
[41] Terzopoulos, IEEE Trans Patt Anal Mach Intell 8 pp 413– (1986)
[42] Terzopoulos, IEEE Trans Patt Anal Mach Intell 10 pp 417– (1988)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.