Updating preconditioners for nonlinear deblurring and denoising image restoration☆

Image restoration is the recovery of images that have been degraded by blur and noise. Nonlinear degenerate diffusion partial differential equation models for image restoration require often the solution of a challenging discrete problem. We consider solving the related discrete models in the time-scale steps by Krylov iterative solvers accelerated by updating a preconditioner based on incomplete factorizations which presents a global computational cost slightly more than linear in the number of the image pixels. We demonstrate the efficiency of the strategy by denoising and deblurring some images with a generalized Alvarez–Lions–Morel-like partial differential equation model discretized by a semi implicit complementary volume scheme.