Advertisement

Journal of Scientific Computing

, Volume 46, Issue 1, pp 20–46 | Cite as

A Unified Primal-Dual Algorithm Framework Based on Bregman Iteration

  • Xiaoqun ZhangEmail author
  • Martin Burger
  • Stanley Osher
Open Access
Article

Abstract

In this paper, we propose a unified primal-dual algorithm framework for two classes of problems that arise from various signal and image processing applications. We also show the connections to existing methods, in particular Bregman iteration (Osher et al., Multiscale Model. Simul. 4(2):460–489, 2005) based methods, such as linearized Bregman (Osher et al., Commun. Math. Sci. 8(1):93–111, 2010; Cai et al., SIAM J. Imag. Sci. 2(1):226–252, 2009, CAM Report 09-28, UCLA, March 2009; Yin, CAAM Report, Rice University, 2009) and split Bregman (Goldstein and Osher, SIAM J. Imag. Sci., 2, 2009). The convergence of the general algorithm framework is proved under mild assumptions. The applications to 1 basis pursuit, TV−L 2 minimization and matrix completion are demonstrated. Finally, the numerical examples show the algorithms proposed are easy to implement, efficient, stable and flexible enough to cover a wide variety of applications.

Keywords

Saddle point Bregman iteration 1 minimization Inexact Uzawa methods Proximal point iteration 

References

  1. 1.
    Arrow, K.J., Hurwicz, L., Uzawa, H.: Studies in Linear and Non-Linear Programming. Stanford University Press, Stanford (1958) zbMATHGoogle Scholar
  2. 2.
    Avinash, C., Malcolm, S.: Principles of Computerized Tomographic Imaging. Society for Industrial and Applied Mathematics, Philadelphia (2001) Google Scholar
  3. 3.
    Beck, A., Teboulle, M.: A fast iterative shrinkage-thresholding algorithm with application to wavelet-based image deblurring. In: IEEE International Conference on Acoustics, Speech, and Signal Processing, pp. 693–696 (2009) Google Scholar
  4. 4.
    Becker, S., Bobin, J., Candès, E.: Nesta: a fast and accurate first-order method for sparse recovery (2009) Google Scholar
  5. 5.
    Borghi, A., Darbon, J., Peyronnet, S., Chan, T.F., Osher, S.: A simple compressive sensing algorithm for parallel many-core architectures. CAM Report 08-64, UCLA (2008) Google Scholar
  6. 6.
    Bramble, J.H., Pasciak, J.E., Vassilev, A.T.: Analysis of the inexact Uzawa algorithm for saddle point problems. SIAM J. Numer. Anal. 34(3), 1072–1092 (1997) zbMATHCrossRefMathSciNetGoogle Scholar
  7. 7.
    Bregman, L.: The relaxation method of finding the common points 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.
    Cai, J.-F., Candès, E.J., Shen, Z.: A singular value thresholding algorithm for matrix completion. SIAM J. Optim. 20(4), 1956–1982 (2010) zbMATHCrossRefMathSciNetGoogle Scholar
  9. 9.
    Cai, J.-F., Osher, S., Shen, Z.: Convergence of the linearized Bregman iteration for 1-norm minimization. Math. Comput. 78, 2127–2136 (2009) zbMATHCrossRefMathSciNetGoogle Scholar
  10. 10.
    Cai, J.-F., Osher, S., Shen, Z.: Split Bregman method and frame based image restoration. CAM Report 09-28, UCLA, March 2009 Google Scholar
  11. 11.
    Cai, J.-F., Osher, S., Shen, Z.: Linearized Bregman iterations for frame-based image deblurring. SIAM J. Imag. Sci. 2(1), 226–252 (2009) zbMATHCrossRefMathSciNetGoogle Scholar
  12. 12.
    Candès, E.J., Recht, B.: Exact matrix completion via convex optimization. Found. Comput. Math. (2008, to appear) Google Scholar
  13. 13.
    Candès, E.J., Romberg, J., Tao, T.: Robust uncertainty principles: exact signal reconstruction from highly incomplete frequency information. IEEE Trans. Inf. Theory 52(2), 489–509 (2006) CrossRefGoogle Scholar
  14. 14.
    Censor, Y., Lent, A.: An iterative row-action method for interval convex programming. J. Optim. Theory Appl. 34(3), 321–353 (1981) zbMATHCrossRefMathSciNetGoogle Scholar
  15. 15.
    Chambolle, A.: An algorithm for total variation minimization and applications. J. Math. Imag. Vis. 20, 89–97 (2004) CrossRefMathSciNetGoogle Scholar
  16. 16.
    Chan, T.F., Golub, G.H., Mulet, P.: A nonlinear primal-dual method for total variation-based image restoration. Math. Oper. Res. 219, 241–252 (1996) MathSciNetGoogle Scholar
  17. 17.
    Chen, G., Teboulle, M.: Convergence analysis of a proximal like minimization algorithm using Bregman functions. SIAM J. Optim., 3 (1993) Google Scholar
  18. 18.
    Chen, G., Teboulle, M.: A proximal-based decomposition method for convex minimization problems. Math. Program. 64(1), 81–101 (1994) zbMATHCrossRefMathSciNetGoogle Scholar
  19. 19.
    Chen, S.S., Donoho, David L., Saunders, M.A.: Atomic decomposition by basis pursuit. SIAM J. Sci. Comput. 20(1), 33–61 (1998) CrossRefMathSciNetGoogle Scholar
  20. 20.
    Combettes, P.L., Wajs, V.R.: Signal recovery by proximal forward-backward splitting. Multiscale Model. Simul. 4(4) (2005) Google Scholar
  21. 21.
    Darbon, J., Osher, S.: Fast discrete optimization for sparse approximations and deconvolutions. Preprint, UCLA (2007) Google Scholar
  22. 22.
    Daubechies, I., Defrise, M., De Mol, C.: An iterative thresholding algorithm for linear inverse problems with a sparsity constraint. Commun. Pure Appl. Math. 57(11), 1413–1457 (2004) zbMATHCrossRefGoogle Scholar
  23. 23.
    Donoho, D.L.: Compressed sensing. IEEE. Trans. Inf. Theory 52, 1289–1306 (2006) CrossRefMathSciNetGoogle Scholar
  24. 24.
    Douglas, J., Rachford, H.H.: On the numerical solutions of heat conduction problems in two and three space variables. Trans. Am. Math. Soc. 1, 97–116 (1976) Google Scholar
  25. 25.
    Eckstein, J., Bertsekas, D.P.: On the Douglas-Rachford splitting method and the proximal point algorithm for maximal monotone operators. Math. Program. 55, 293–318 (1992) zbMATHCrossRefMathSciNetGoogle Scholar
  26. 26.
    Esser, E.: Applications of Lagrangian-based alternating direction methods and connections to split Bregman. CAM Report 09-31, UCLA, April 2009 Google Scholar
  27. 27.
    Esser, E., Zhang, X., Chan, T.: A general framework for a class of first order primal-dual algorithms for TV minimization. CAM Report 09-67, UCLA, August 2009 Google Scholar
  28. 28.
    Figueiredo, M.A.T., Nowak, R.D., Wright, S.J.: Gradient projection for sparse reconstruction: Application to compressed sensing and other inverse problems. IEEE J. Sel. Top. Signal Proces. 1(4), 586–597 (2007) CrossRefGoogle Scholar
  29. 29.
    Gabay, D.: Applications of the method of multipliers to variational inequalities. In: Augmented Lagrangian Methods: Applications to the Solution of Boundary-Value Problems. North-Holland, Amsterdam (1983) Google Scholar
  30. 30.
    Glowinski, R., Le Tallec, P.: Augmented Lagrangian and Operator Splitting Methods in Nonlinear Mechanics. SIAM, Philadelphia (1989) zbMATHGoogle Scholar
  31. 31.
    Goldstein, T., Bresson, X., Osher, S.: Geometric applications of the split Bregman method: segmentation and surface reconstruction. UCLA CAM Report, 09-06, February (2009) Google Scholar
  32. 32.
    Goldstein, T., Osher, S.: The split Bregman method for l 1 regularized problems. SIAM J. Imag. Sci., 2 (2009) Google Scholar
  33. 33.
    Guo, X., Li, F., Michael, K.Ng.: A fast L1-TV algorithm for image restoration. ICM Technical Report 08-13, Hong Kong Baptist University, November 2008 Google Scholar
  34. 34.
    Hale, E.T., Yin, W., Zhang, Y.: A fixed-point continuation method for l1-regularized minimization with applications to compressed sensing. SIAM J. Optim. 19, 1107–1130 (2008) zbMATHCrossRefMathSciNetGoogle Scholar
  35. 35.
    Hestenes, M.R.: Multiplier and gradient methods. J. Optim. Theory Appl. 4, 303–320 (1969) zbMATHCrossRefMathSciNetGoogle Scholar
  36. 36.
    Kim, S., Koh, K., Lustig, M., Boyd, S., Gorinevsky, D.: An interior-point method for large-scale l 1-regularized least squares. IEEE J. Sel. Top. Signal Proces. 1(4), 606–617 (2007) CrossRefGoogle Scholar
  37. 37.
    Lemaréchal, C., Sagastizábal, C.: Practical aspects of the Moreau-Yosida regularization, I: Theoretical properties (1994) Google Scholar
  38. 38.
    Li, Y., Osher, S.: Coordinate descent optimization for 1 minimization with application to compressed sensing; a greedy algorithm. Inverse Probl. Imaging 3(3), 487–503 (2009) zbMATHCrossRefMathSciNetGoogle Scholar
  39. 39.
    Ma, S., Goldfarb, D., Chen, L.: Fixed point and Bregman iterative methods for matrix rank minimization. Technical Report, Columbia University (2008) Google Scholar
  40. 40.
    Moreau, J.-J.: Fonctions convexes duales et points proximaux dans un espace hilbertien. C. R. Acad. Sci. Paris Ser. A Math. 255, 2897–2899 (1962) zbMATHMathSciNetGoogle Scholar
  41. 41.
    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) zbMATHCrossRefMathSciNetGoogle Scholar
  42. 42.
    Osher, S., Mao, Y., Dong, B., Yin, W.: Fast linearized Bregman iteration for compressive sensing and sparse denoising. Commun. Math. Sci. 8(1), 93–111 (2010) zbMATHMathSciNetGoogle Scholar
  43. 43.
    Powell, M.J.D.: A method for nonlinear constraints in minimization problems. Optimization (1972) Google Scholar
  44. 44.
    Rockafellar, R.T.: Augmented Lagrangians and applications of the proximal point algorithm in convex programming. Math. Oper. Res. 82, 421–439 (1956) Google Scholar
  45. 45.
    Rudin, L.I., Osher, S., Fatemi, E.: Nonlinear total variation based noise removal algorithms. Physica D 60, 259–268 (1992) zbMATHGoogle Scholar
  46. 46.
    Setzer, S.: Split Bregman algorithm, Douglas-Rachford splitting and frame shrinkage. In: Morken, K., Lysaker, M., Lie, K.-A., Tai, X.-C. (eds.) Scale Space and Variational Methods in Computer Vision. LNCS, vol. 5567, pp. 464–476. Springer, Berlin (2009) CrossRefGoogle Scholar
  47. 47.
    Tai, X.-C., Wu, C.: Augmented Lagrangian method, dual methods and split Bregman iteration for ROF model. CAM Report 09-05, UCLA, January 2009 Google Scholar
  48. 48.
    Yang, J., Zhang, Y., Yin, W.: An efficient TV-L1 algorithm for deblurring multichannel images corrupted by impulsive noise. SIAM J. Sci. Comput. 31(4) (2009) Google Scholar
  49. 49.
    Yin, W.: Analysis and generalizations of the linearized Bregman method. CAAM Report, Rice University (2009) Google Scholar
  50. 50.
    Yin, W., Osher, S., Goldfarb, D., Darbon, J.: Bregman iterative algorithms for 1 minimization with applications to compressed sensing. SIAM J. Imag. Sci. 1, 143–168 (2008) zbMATHCrossRefMathSciNetGoogle Scholar
  51. 51.
    Zhang, X., Burger, M., Bresson, X., Osher, S.: Bregmanized nonlocal regularization for deconvolution and sparse reconstruction. CAM Report 09-03, UCLA, January 2009 Google Scholar
  52. 52.
    Zhu, M., Chan, T.: An efficient primal-dual hybrid gradient algorithm for total variation image restoration. CAM Report 08-34, UCLA, May 2008 Google Scholar

Copyright information

© The Author(s) 2010

Open AccessThis is an open access article distributed under the terms of the Creative Commons Attribution Noncommercial License (https://creativecommons.org/licenses/by-nc/2.0), which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.

Authors and Affiliations

  1. 1.Department of MathematicsShanghai Jiao Tong UniversityShanghaiP.R. China
  2. 2.Institute for Computational and Applied MathematicsWestfälische Wilhelms-UniversitätMünsterGermany
  3. 3.Department of MathematicsUCLALos AngelesUSA

Personalised recommendations