Image restoration with regularization convex optimization approach

Document Type : Researsh Articles

Authors

• Abdolreza Rashno
• Foroogh Sadat Tabataba
• Saeed Sadri

Isfahan University of Technology

10.22067/ess.v2i2.42760

Abstract

In this paper, Tikhonov regularization with l-curve parameter estimation as convex optimization problem has been proposed in image restoration as a solution of ill-posed problem stem from sparse and large scale blurring matrix which has many singular values of different orders of magnitude close to the origin. Also, since the restored image is so sensitive to initial guess (start point) of optimization algorithm, a new schema for feasible set and feasible start point has been proposed. Some numerical results show the efficiency of the proposed algorithm in comparison with previous proposed methods.

Keywords

• Image restoration
• convex optimization
• Tikhonov regularization
• l-curve estimation

---

References

[1] A. Bouhamidi, and K. Jbilou, “Sylvester Tikhonov-regularization methods in image restoration,”Journal of Computational and Applied Mathematics, vol. 206, no. 1, pp. 86-98, 2007.

[2] A. Bouhamidi, K. Jbilou, and M. Raydan, “Convex constrained optimization for large-scale ill-conditioned generalized Sylvester equations,” Computational Optimization and Applications, vol. 48, No. 2, pp. 233-253, 2011.

[3] S. Morigi, L. Reichel, and F. Sgallari, “An interior-point method for large constrained discrete ill-possed problems,”Journal of Computational and Applied Mathematics, vol. 233, pp. 1288-1297, 20010.

[4] J. Nagy and Z. Strakos, “Enforcing nonnegativity in image reconstruction algorithms,” in: David C. Wilson, et al. (Eds.), Mathematical Modeling, Estimation, and Imaging, Proceedings of SPIE, vol. 4121, pp. 182-190, 2000.

[5] M. Rojas, and T. Steihaug, “An interior-point trust-region-based method for large-scale non-negative regularization,” Inverse Problems, vol. 18, pp. 1291-1307, 2002.

[6] D. Calvetti, S. Morigi, L. Reichel, and F. Sgallari, “Tikhonov regularization and the L-curve for large discrete ill-posed problems,” Journal of Computational and Applied Mathematics, vol. 123, pp. 423-446, 2000.

[7] A. Bouhamidi, R. Enkhbat, and K. Jbilou, “Conditional gradient Tikhonov method for a convex optimization problem in image restoration,” Journal of Computational and Applied Mathematics, vol. 255, pp. 580-592, 2014.

[8] K. Bredies, K. Kunisch, and T. Pock, “Total generalized variation,” SIAM J. Imag. Sci,vol. 3, pp. 492-526, 2010.

[9] W. Zhu and T.F. Chan, “Image denoising using mean curvature,”SIAM J. Imag. Sci.,vol. 5,pp. 1-32, 2012.

[10] J. Liu, T. Z. Huang,I. W. Selesnick, X. G. Lv, and P. Y. Chen, “Image restoration using total variation with overlapping group sparsity,”Information Sciences,vol. 295, pp. 232-246, 2015.

[11] J. F. Cai, S. Osher, and Z. W. Shen, “Linearized bregman iterations for frame-based image deblurring,”SIAM J. Imag. Sci., vol. 2, pp. 226-252, 2009.

[12] L. Sun, and K. Chen, “A new iterative algorithm for mean curvature-based variational image denoising,” BIT Numerical Mathematics, vol. 54, issue 2, pp. 523-553, 2013.

[13] B. Morini, M. Porcelli, and R. Chan, “A reduced Newton method for constrained linear least-squares problems,” Journal of Computational and Applied Mathematics, vol. 233, pp. 2200-2212, 2010.