Performance comparison of iterative reweighting methods for total variation regularization

Iteratively Reweighted Least Squares (IRLS) is a well-established method of optimizing ℓp norm problems such as Total Variation (TV) regularization. Within this general framework, there are several possible ways of constructing the weights and the form of the linear system that is iteratively solved as part of the algorithm. Many of these choices are equally reasonable from a theoretical perspective, and there has, thus far, been no systematic comparison between them. In this paper we provide such a comparison between the main choices in IRLS algorithms for ℓ1- and ℓ2-TV denoising, finding that there is a significant variation in the computational cost and reconstruction quality of the different variants.

[1]  Mila Nikolova,et al.  Regularizing Flows for Constrained Matrix-Valued Images , 2004, Journal of Mathematical Imaging and Vision.

[2]  B. Mercier,et al.  A dual algorithm for the solution of nonlinear variational problems via finite element approximation , 1976 .

[3]  Thierry Blu,et al.  Monte-Carlo Sure: A Black-Box Optimization of Regularization Parameters for General Denoising Algorithms , 2008, IEEE Transactions on Image Processing.

[4]  L. Rudin,et al.  Nonlinear total variation based noise removal algorithms , 1992 .

[5]  Tony F. Chan,et al.  Spatially and Scale Adaptive Total Variation Based Regularization and Anisotropic Diiusion in Image Processing , 1996 .

[6]  Paul A. Rodríguez,et al.  Total Variation Regularization Algorithms for Images Corrupted with Different Noise Models: A Review , 2013, J. Electr. Comput. Eng..

[7]  Elwood T. Olsen,et al.  L1 and L∞ minimization via a variant of Karmarkar's algorithm , 1989, IEEE Trans. Acoust. Speech Signal Process..

[8]  Brendt Wohlberg,et al.  A comparison of the computational performance of Iteratively Reweighted Least Squares and alternating minimization algorithms for ℓ1 inverse problems , 2012, 2012 19th IEEE International Conference on Image Processing.

[9]  Mila Nikolova,et al.  Minimizers of Cost-Functions Involving Nonsmooth Data-Fidelity Terms. Application to the Processing of Outliers , 2002, SIAM J. Numer. Anal..

[10]  Paul Rodríguez,et al.  Spatially adaptive Total Variation image denoising under salt and pepper noise , 2011, 2011 19th European Signal Processing Conference.

[11]  Brendt Wohlberg,et al.  UPRE method for total variation parameter selection , 2010, Signal Process..

[12]  C. Vogel.  Computational Methods for Inverse Problems , 1987 .

[13]  R. Glowinski,et al.  Sur l'approximation, par éléments finis d'ordre un, et la résolution, par pénalisation-dualité d'une classe de problèmes de Dirichlet non linéaires , 1975 .

[14]  Nahum Kiryati,et al.  Image Deblurring in the Presence of Impulsive Noise , 2006, International Journal of Computer Vision.

[15]  Brendt Wohlberg,et al.  Efficient Minimization Method for a Generalized Total Variation Functional , 2009, IEEE Transactions on Image Processing.

[16]  Stephen P. Boyd,et al.  Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers , 2011, Found. Trends Mach. Learn..

[17]  Raymond H. Chan,et al.  Salt-and-pepper noise removal by median-type noise detectors and detail-preserving regularization , 2005, IEEE Transactions on Image Processing.

[18]  Curtis R. Vogel,et al.  Ieee Transactions on Image Processing Fast, Robust Total Variation{based Reconstruction of Noisy, Blurred Images , 2022 .

[19]  Tom Goldstein,et al.  The Split Bregman Method for L1-Regularized Problems , 2009, SIAM J. Imaging Sci..

[20]  Yiqiu Dong,et al.  Automated Regularization Parameter Selection in Multi-Scale Total Variation Models for Image Restoration , 2011, Journal of Mathematical Imaging and Vision.

[21]  Robert D. Nowak,et al.  On Total Variation Denoising: A New Majorization-Minimization Algorithm and an Experimental Comparisonwith Wavalet Denoising , 2006, 2006 International Conference on Image Processing.