Scale-Space and Edge Detection Using Anisotropic Diffusion
IEEE Transactions on Pattern Analysis and Machine Intelligence
SIAM Journal on Scientific Computing
Iterative Methods for Sparse Linear Systems
Iterative Methods for Sparse Linear Systems
An Algorithm for Total Variation Minimization and Applications
Journal of Mathematical Imaging and Vision
On Semismooth Newton's Methods for Total Variation Minimization
Journal of Mathematical Imaging and Vision
Iterative Algorithms Based on Decoupling of Deblurring and Denoising for Image Restoration
SIAM Journal on Scientific Computing
Cascadic Multiresolution Methods for Image Deblurring
SIAM Journal on Imaging Sciences
Adaptive total variation denoising based on difference curvature
Image and Vision Computing
Cascadic multilevel methods for fast nonsymmetric blur- and noise-removal
Applied Numerical Mathematics
Updating preconditioners for nonlinear deblurring and denoising image restoration
Applied Numerical Mathematics
Fast nonconvex nonsmooth minimization methods for image restoration and reconstruction
IEEE Transactions on Image Processing
Tikhonov regularization based on generalized Krylov subspace methods
Applied Numerical Mathematics
Efficient and reliable schemes for nonlinear diffusion filtering
IEEE Transactions on Image Processing
Blind image deconvolution using a banded matrix method
Numerical Algorithms
Hi-index | 7.29 |
Alternating methods for image deblurring and denoising have recently received considerable attention. The simplest of these methods are two-way methods that restore contaminated images by alternating between deblurring and denoising. This paper describes Krylov subspace-based two-way alternating iterative methods that allow the application of regularization operators different from the identity in both the deblurring and the denoising steps. Numerical examples show that this can improve the quality of the computed restorations. The methods are particularly attractive when matrix-vector products with a discrete blurring operator and its transpose can be evaluated rapidly, but the structure of these operators does not allow inexpensive diagonalization.