Multi-parameter Tikhonov regularization — An augmented approach

Abstract

We study multi-parameter regularization (multiple penalties) for solving linear inverse problems to promote simultaneously distinct features of the sought-for objects. We revisit a balancing principle for choosing regularization parameters from the viewpoint of augmented Tikhonov regularization, and derive a new parameter choice strategy called the balanced discrepancy principle. A priori and a posteriori error estimates are provided to theoretically justify the principles, and numerical algorithms for efficiently implementing the principles are also provided. Numerical results on deblurring are presented to illustrate the feasibility of the balanced discrepancy principle.

Keywords

Multi-parameter regularization / Augmented Tikhonov regularization, Balanced discrepancy principle

References

[1] Belge M, Kilmer M E, Miller E L. Efficient determination of multiple regularization parameters in a generalized L-curve framework. Inverse Problems, 2002, 18(4): 1161-1183

[2] Broyden C G. A class of methods for solving nonlinear simultaneous equations. Math. Comp., 1965, 19(92): 577-593

[3] Burger M, Osher S. Convergence rates of convex variational regularization. Inverse Problems, 2004, 20(5): 1411-1420

[4] Chen Z, Lu Y, Xu Y, Yang H. Multi-parameter Tikhonov regularization for linear ill-posed operator equations. J. Comput. Math., 2008, 26(1): 37-55

[5] Engl H W, Hanke M, Neubauer A. Regularization of Inverse Problems, 1996, Dordrecht: Kluwer

[6] Hofmann B, Kaltenbacher B, Poeschl C, Scherzer O. A convergence rates result for Tikhonov regularization in Banach spaces with non-smooth operators. Inverse Problems, 2007, 23(3): 987-1010

[7] Ito K, Jin B, Takeuchi T. Multi-parameter Tikhonov regularization. Methods Appl. Anal., 2011, 18(1): 31-46

[8] Ito K, Jin B, Takeuchi T. A regularization parameter for nonsmooth Tikhonov regularization. SIAM J. Sci. Comput., 2011, 33(3): 1415-1438

[9] Ito K, Kunisch K. BV-type regularization methods for convoluted objects with edge, flat and grey scales. Inverse Problems, 2000, 16(4): 909-928

[10] Ivanov V K, Vasin V V, Tanana V P. Theory of Linear Ill-Posed Problems and Its Applications, 2002 2nd edition Utrecht: VSP

[11] Jin B, Lorenz D A. Heuristic parameter-choice rules for convex variational regularization based on error estimates. SIAM J. Numer. Anal., 2010, 48(3): 1208-1229

[12] Jin B, Zou J. Augmented Tikhonov regularization. Inverse Problems, 2009, 25(2): 025001

[13] Lu S, Pereverzev S V. Multi-parameter regularization and its numerical regularization. Numer. Math., 2011, 118(1): 1-31

[14] Lu S, Pereverzev S V, Shao Y, Tautenhahn U. Discrepancy curves for multi-parameter regularization. J. Inv. Ill-Posed Probl., 2010, 18(6): 655-676

[15] Lu Y, Shen L, Xu Y. Multi-parameter regularization methods for high-resolution image reconstruction with displacement errors. IEEE Trans. Circuits Syst. I. Regul. Pap., 2007, 54(8): 1788-1799

[16] Mathé P. The Lepskii principle revisited. Inverse Problems, 2006, 22(3): L11-L15

[17] Stephanakis I M. Regularized image restoration in multiresolution spaces. Opt. Eng., 1997, 36(6): 1738-1744

[18] Xu P, Fukuda Y, Liu Y. Multiple parameter regularization: numerical solutions and applications to the determination of geopotential from precise satellite orbits. J. Geod., 2006, 80(1): 17-27

[19] Zou H, Hastie T. Regularization and variable selection via the elastic net. J. R. Stat. Soc. Ser. B, 2005, 67(2): 301-320