Journal of Scientific Computing

, Volume 54, Issue 2–3, pp 269–310 | Cite as

Higher-Order TV Methods—Enhancement via Bregman Iteration

  • Martin Benning
  • Christoph Brune
  • Martin Burger
  • Jahn Müller
Article

Abstract

In this work we analyze and compare two recent variational models for image denoising and improve their reconstructions by applying a Bregman iteration strategy. One of the standard techniques in image denoising, the ROF-model (cf. Rudin et al. in Physica D 60:259–268, 1992), is well known for recovering sharp edges of a signal or image, but also for producing staircase-like artifacts. In order to overcome these model-dependent deficiencies, total variation modifications that incorporate higher-order derivatives have been proposed (cf. Chambolle and Lions in Numer. Math. 76:167–188, 1997; Bredies et al. in SIAM J. Imaging Sci. 3(3):492–526, 2010). These models reduce staircasing for reasonable parameter choices. However, the combination of derivatives of different order leads to other undesired side effects, which we shall also highlight in several examples.

The goal of this paper is to analyze capabilities and limitations of the different models and to improve their reconstructions in quality by introducing Bregman iterations. Besides general modeling and analysis we discuss efficient numerical realizations of Bregman iterations and modified versions thereof.

Keywords

Total variation regularization Higher order methods Staircasing Exact solutions Bregman iteration 

Notes

Acknowledgements

Financial support is acknowledged to the German Science Foundation (DFG) via grants SFB 656 (Subproject B2) and BU 2327/1. The third author thanks Stanley Osher for initiating his interest in variational methods for image processing.

References

  1. 1.
    Alter, F., Caselles, V., Chambolle, A.: Evolution of characteristic functions of convex sets in the plane by the minimizing total variation flow. Interfaces Free Bound. 7(1), 29–53 (2005) MathSciNetMATHCrossRefGoogle Scholar
  2. 2.
    Bauschke, H.H., Combettes, P.L.: Convex Analysis and Monotone Operator Theory in Hilbert Spaces. CMS Books in Mathematics/Ouvrages de Mathématiques de la SMC. Springer, New York (2011). With a foreword by Hédy Attouch MATHCrossRefGoogle Scholar
  3. 3.
    Benning, M.: Singular regularization of inverse problems. Ph.D. thesis, University of Münster, Institute for Computational and Applied Mathematics, Einsteinstr. 62, 48149 Münster, May 2011 Google Scholar
  4. 4.
    Bertozzi, A.L., Greer, J.B., Osher, S., Vixie, K.: Nonlinear regularizations of TV based PDEs for image processing. In: Chen, G.-Q., Gasper, G., Jerome, J. (eds.) AMS Series of Contemporary Mathematics, vol. 371, pp. 29–40 (2005) Google Scholar
  5. 5.
    Bishop, R.L., Goldberg, S.I.: Tensor Analysis on Manifolds. Dover, New York (1980) Google Scholar
  6. 6.
    Bredies, K., Kunisch, K., Pock, T.: Total generalized variation. SIAM J. Imaging Sci. 3(3), 492–526 (2010) MathSciNetMATHCrossRefGoogle Scholar
  7. 7.
    Bregman, L.M.: The relaxation method for finding the common point of convex sets and its application to the solution of problems in convex programming. USSR Comput. Math. Math. Phys. 7, 200–217 (1967) CrossRefGoogle Scholar
  8. 8.
    Buades, A., Coll, B., Morel, J.M.: A review of image denoising algorithms, with a new one. Multiscale Model. Simul. 4(2), 490–530 (2005) MathSciNetMATHCrossRefGoogle Scholar
  9. 9.
    Burger, M., Osher, S.: Convergence rates of convex variational regularization. Inverse Probl. 20, 1411–1421 (2004) MathSciNetMATHCrossRefGoogle Scholar
  10. 10.
    Burger, M., Osher, S.: A guide to tv zoo. In: Level Set and PDE-based Reconstruction Methods. Springer, Berlin (2012, to appear) Google Scholar
  11. 11.
    Burger, M., Gilboa, G., Osher, S., Xu, J.: Nonlinear inverse scale space methods. Commun. Math. Sci. 4, 179–212 (2006) MathSciNetMATHGoogle Scholar
  12. 12.
    Burger, M., Frick, K., Osher, S., Scherzer, O.: Inverse total variation flow. Multiscale Model. Simul. 6(2), 366–395 (2007) MathSciNetMATHCrossRefGoogle Scholar
  13. 13.
    Caselles, V., Chambolle, A., Novaga, M.: The discontinuity set of solutions of the TV denoising problem and some extensions. Multiscale Model. Simul. 6(3), 879–894 (2007) MathSciNetMATHCrossRefGoogle Scholar
  14. 14.
    Chambolle, A.: An algorithm for total variation minimization and applications. J. Math. Imaging Vis. 20(1–2), 89–97 (2004). Special issue on mathematics and image analysis MathSciNetGoogle Scholar
  15. 15.
    Chambolle, A., Lions, P.-L.: Image recovery via total variation minimization and related problems. Numer. Math. 76, 167–188 (1997) MathSciNetMATHCrossRefGoogle Scholar
  16. 16.
    Chambolle, A., Pock, T.: A first-order primal-dual algorithm for convex problems with applications to imaging. J. Math. Imaging Vis. 40(1), 120–145 (2011) MathSciNetMATHCrossRefGoogle Scholar
  17. 17.
    Chan, T.F., Shen, J.: Image Processing and Analysis: Variational, PDE, Wavelet and Stochastic Methods. SIAM, Philadelphia (2005) MATHCrossRefGoogle Scholar
  18. 18.
    Chan, T., Marquina, A., Mulet, P.: High-order total variation-based image restoration. SIAM J. Sci. Comput. 22(2), 503–516 (2000) MathSciNetMATHCrossRefGoogle Scholar
  19. 19.
    Esser, E., Zhang, X., Chan, T.F.: A general framework for a class of first order primal-dual algorithms for convex optimization in imaging science. SIAM J. Imaging Sci. 3(4), 1015–1046 (2010) MathSciNetMATHCrossRefGoogle Scholar
  20. 20.
    Gilboa, G., Osher, S.: Nonlocal operators with applications to image processing. Multiscale Model. Simul. 7(3), 1005–1028 (2008) MathSciNetMATHCrossRefGoogle Scholar
  21. 21.
    Goldstein, T., Osher, S.: The split Bregman method for l1-regularized problems. SIAM J. Imaging Sci. 2(2), 323–343 (2009) MathSciNetMATHCrossRefGoogle Scholar
  22. 22.
    Hinze, M., Pinnau, R., Ulbrich, M., Ulbrich, S.: Optimization with PDE Constraints. Mathematical Modelling: Theory and Applications, vol. 23. Springer, New York (2009) MATHCrossRefGoogle Scholar
  23. 23.
    Kiwiel, K.C.: Proximal minimization methods with generalized Bregman functions. SIAM J. Control Optim. 35(4), 1142–1168 (1997) MathSciNetMATHCrossRefGoogle Scholar
  24. 24.
    Knoll, F., Bredies, K., Pock, T., Stollberger, R.: Second order total generalized variation (TGV) for MRI. Magn. Reson. Med. 65(2), 480–491 (2011) CrossRefGoogle Scholar
  25. 25.
    Meyer, Y.: Oscillating Patterns in Image Processing and Nonlinear Evolution Equations: The Fifteenth Dean Jacqueline B. Lewis Memorial Lectures. University Lecture Series, vol. 22. Am. Math. Soc., Boston (2001) MATHGoogle Scholar
  26. 26.
    Osher, S., Burger, M., Goldfarb, D., Xu, J., Yin, W.: An iterative regularization method for total variation-based image restoration. Multiscale Model. Simul. 4(2), 460–489 (2005) MathSciNetMATHCrossRefGoogle Scholar
  27. 27.
    Rudin, L.I., Osher, S., Fatemi, E.: Nonlinear total variation based noise removal algorithms. Physica D 60, 259–268 (1992) MATHCrossRefGoogle Scholar
  28. 28.
    Scherzer, O.: Denoising with higher order derivatives of bounded variation and an application to parameter estimation. Computing 60(1), 1–27 (1998) MathSciNetMATHCrossRefGoogle Scholar
  29. 29.
    Setzer, S., Steidl, G.: Variational methods with higher order derivatives in image processing. In: Neamtu, M., Schumaker, L.L. (eds.) Approximation XII. Nashboro Press, Brentwood (2008) Google Scholar
  30. 30.
    Setzer, S., Steidl, G., Teuber, T.: Infimal convolution regularizations with discrete l1-type functionals. Commun. Math. Sci. 9(3), 797–827 (2011) MathSciNetGoogle Scholar
  31. 31.
    Stefan, W., Renaut, R.A., Gelb, A.: Improved total variation-type regularization using higher order edge detectors. SIAM J. Imaging Sci. 3(2), 232–251 (2010) MathSciNetMATHCrossRefGoogle Scholar
  32. 32.
    Strong, D., Chan, T.: Edge-preserving and scale-dependent properties of total variation regularization. Inverse Probl. 19(6), S165–S187 (2003). Special section on imaging MathSciNetMATHCrossRefGoogle Scholar
  33. 33.
    Zhang, X., Burger, M., Osher, S.: A unified primal-dual algorithm framework based on Bregman iteration. J. Sci. Comput. 46(1), 20–46 (2011) MathSciNetMATHCrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media New York 2012

Authors and Affiliations

  • Martin Benning
    • 1
  • Christoph Brune
    • 2
  • Martin Burger
    • 1
  • Jahn Müller
    • 1
  1. 1.Institut für Numerische und Angewandte MathematikWestfälische Wilhelms-Universität MünsterMünsterGermany
  2. 2.Department of MathematicsUniversity of California Los AngelesLos AngelesUSA

Personalised recommendations