Iterative methods for anisotropic diffusion of speckled medical images

This paper deals with filtering multiplicative noise that corrupts ultrasound medical images. The aim is to increase their contrast, while preserving edges and structures. From a mathematical point of view, this requires solving numerically an evolutive nonlinear boundary value problem. Several discretization schemes are studied for the stationary problem as well as for the evolutive problem. The behavior of various iterative methods is analyzed when semi-implicit time marching schemes are used. Results of sequential simulations are presented.