Domain decomposition methods in image denoising using Gaussian curvature

摘要

A domain decomposition method is used to solve a Gaussian curvature driven flow to denoise digital images. The algorithm is embarrassingly parallel and is ideal for parallel computers. We use the scheme of Lee–Keun [Noise removal based on nonlinear diffusion with Gauss curvature conductance, 2003, Preprint] in which the diffusion coefficient contains a term involving the Gaussian curvature. This minimizes the occurrence of patches and tips of cones are not flattened.