Hostname: page-component-7c8c6479df-p566r Total loading time: 0 Render date: 2024-03-29T01:41:35.726Z Has data issue: false hasContentIssue false

A Convex and Exact Approach to Discrete Constrained TV-L1 Image Approximation

Published online by Cambridge University Press:  28 May 2015

Jing Yuan*
Affiliation:
Computer Science Department, Middlesex College, University of Western Ontario, London, Ontario, N6A 5B7, UK
Juan Shi*
Affiliation:
Division of Mathematical Sciences, School of Physical and Mathematical Sciences, Nanyang Technological University, Singapore
Xue-Cheng Tai*
Affiliation:
Division of Mathematical Sciences, School of Physical and Mathematical Sciences, Nanyang Technological University, Singapore Department of Mathematics, University of Bergen, Norway
*
Corresponding author. Email: cn.yuanjing@gmail.com
Corresponding author. Email: shij0004@e.ntu.edu.sg
Corresponding author. Email: tai@mi.uib.no
Get access

Abstract

We study the TV-L1 image approximation model from primal and dual perspective, based on a proposed equivalent convex formulations. More specifically, we apply a convex TV-L1 based approach to globally solve the discrete constrained optimization problem of image approximation, where the unknown image function u(x) ∈ {f1,…,fn}, ∀x ∈ Ω. We show that the TV-L1 formulation does provide an exact convex relaxation model to the non-convex optimization problem considered. This result greatly extends recent studies of Chan et al., from the simplest binary constrained case to the general gray-value constrained case, through the proposed rounding scheme. In addition, we construct a fast multiplier-based algorithm based on the proposed primal-dual model, which properly avoids variability of the concerning TV-L1 energy function. Numerical experiments validate the theoretical results and show that the proposed algorithm is reliable and effective.

Type
Research Article
Copyright
Copyright © Global-Science Press 2011

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

[1]Alliney, S.. Digital filters as absolute norm regularizers. IEEE Trans. Signal Process., 40(6):15481562, June 1992.Google Scholar
[2]Alliney, S.. Recursive median filters of increasing order: a variational approach. IEEE Trans. Signal Process., 44(6):13461354, June 1996.Google Scholar
[3]Bae, E., Yuan, J., Tai, X.-C., and Boykov, Y.. A study on continuous max-flow and min-cut approaches part ii: Multiple linearly ordered labels. Technical report CAM-10-62, UCLA, CAM, 2010.Google Scholar
[4]Bae, E., Yuan, J., and Tai, X.C.. Global minimization for continuous multiphase partitioning problems using a dual approach. Technical report CAM09-75, UCLA, CAM, September 2009.Google Scholar
[5]Bresson, X., Esedoglu, S., Vandergheynst, P., Thiran, J.P., and Osher, S.. Fast global minimization of the active contour/snake model. J. Math. Imaging Vision, 28(2):151167, 2007.Google Scholar
[6]Burger, M., Osher, S., Xu, J., and Gilboa, G.. Nonlinear inverse scale space methods for image restoration. Lect. Notes Comput. Sci., 3752:25, 2005.Google Scholar
[7]Chambolle, A.. An algorithm for total variation minimization and applications. J. Math. Imaging Vision, 20(1-2):8997, Jan 2004.Google Scholar
[8]Chan, T.F. and Esedoḡlu, S.. Aspects of total variation regularized L 1 function approximation. SIAM J. Appl. Math., 65(5):18171837 (electronic), 2005.Google Scholar
[9]Chan, T.F., Esedoglu, S., and Nikolova, M.. Algorithms for finding global minimizers of image segmentation and denoising models. SIAM J. Appl. Math., 66(5):16321648 (electronic), 2006.Google Scholar
[10]Darbon, J. and Sigelle, M.. Image restoration with discrete constrained total variation part i: Fast and exact optimization. J. Math. Imaging Vision, 26(3):261276, 2006.Google Scholar
[11]Darbon, J. and Sigelle, M.. Image restoration with discrete constrained total variation part ii: Levelable functions, convex priors and non-convex cases. J. Math. Imaging Vision, 26(3):277291, 2006.Google Scholar
[12]Dong, Y., Hintermuller, M., and Neri, M.. An Efficient Primal-Dual Method for L1 TV Image Restoration. SIAM J. Imaging Sci., 2(4):11681189, 2009.Google Scholar
[13]Ekeland, I. and Téman, R.. Convex analysis and variational problems. Society for Industrial and Applied Mathematics, Philadelphia, PA, USA, 1999.Google Scholar
[14]Esser, E.. Applications of lagrangian-based alternating direction methods and connections to split bregman. Technical report CAM-09-31, UCLA, CAM, 2009.Google Scholar
[15]Fan, K.. Minimax theorems. Proc. Nat. Acad. Sci. U. S. A., 39:4247, 1953.Google Scholar
[16]Giusti, E.. Minimal surfaces and functions of bounded variation. Australian National University, Canberra, 1977.Google Scholar
[17]Hintermuller, M., Ito, K., and Kunisch, K.. The primal-dual active set strategy as a semismooth Newton method. SIAM J. Optim., 13:865, 2002.Google Scholar
[18]Huang, Y., Ng, M.K., and Wen, Y.W.. A fast total variation minimization method for image restoration. Multiscale Model. Simul., 7:774, 2008.Google Scholar
[19]Meyer, Y.. Oscillating patterns in image processing and nonlinear evolution equations, volume 22 of University Lecture Series. American Mathematical Society, Providence, RI, 2001. The fifteenth Dean Jacqueline B. Lewis memorial lectures.Google Scholar
[20]Osher, S., Burger, M., Goldfarb, D., Xu, J., and Yin, W.. An iterative regularization method for total variation-based image restoration. Multiscale Model. Simul., 4(2):460489, 2006.Google Scholar
[21]Paragios, N., Chen, Y., and Faugeras, O.. Handbook of Mathematical Models in Computer Vision. Springer-Verlag New York, Inc., Secaucus, NJ, USA, 2005.Google Scholar
[22]Pock, T., Schoenemann, T., Graber, G., Bischof, H., and Cremers, D.. A convex formulation of continuous multi-label problems. In ECCV '08: Proceedings of the 10th European Conference on Computer Vision, pages 792805, Berlin, Heidelberg, 2008. Springer-Verlag.Google Scholar
[23]Rockafellar, R.T.. Convex analysis. Princeton Mathematical Series, No. 28. Princeton University Press, Princeton, N.J., 1970.Google Scholar
[24]Rudin, L., Osher, S., and Fatemi, E.. Nonlinear total variation based noise removal algorithms. Physica D, 60(1-4):259268, 1992.Google Scholar
[25]Setzer, S.. Split Bregman algorithm, Douglas-Rachford splitting and frame shrinkage, pages 464476. Springer, LNCS vol. 5567, 2009.Google Scholar
[26]Setzer, S., Steidl, G., and Teuber, T.. Deblurring Poissonian images by split Bregman techniques. J. Visual Commun. Image Repres., 2009. accepted.Google Scholar
[27]Steidl, G., Weickert, J., Brox, T., Mrázek, P. and Welk, M.. On the equivalence of soft wavelet shrinkage, total variation diffusion, total variation regularization, and sides. SIAM J. Numer. Anal, 42:686713, 2004.Google Scholar
[28]Tai, X.C. and Wu, C.. Augmented Lagrangian Method, Dual Methods, and Split Bregman Iteration for ROF Model, pages 502513. Springer, LNCS vol. 5567, 2009.Google Scholar
[29]Wu, C., Zhang, J., and Tai, X.C.. Augmented lagrangian method for total variation restoration with non-quadratic fidelity. Technical report CAM-09-82, UCLA, CAM, 2009.Google Scholar
[30]Yin, W., Goldfarb, D., and Osher, S.. The total variation regularized l 1 model for multiscale decomposition. Multiscale Model. Simul., 6(1):190211, 2007.Google Scholar
[31]Zhang, X., Burger, M., and Osher, S.. A unified primal-dual algorithm framework based on bregman iteration. Technical report CAM-09-99, UCLA, CAM, 2009.Google Scholar