Nonlinear Hyperbolic PDE-based Filter for Mixed Poisson-Gaussian Noise Removal from X-ray Images

A nonlinear partial differential equation (PDE) – based mixed noise reduction technique for medical X-Ray images is proposed in this paper. The filtering approach uses a novel well-posed second-order hyperbolic PDE model that is solved numerically applying a fast-converging finite difference method-based discretization algorithm. This iterative numerical approximation scheme restores successfully the medical images affected by mixtures of noise, since it removes properly both the Gaussian and Poisson noise. The effectiveness of the introduced technique is illustrated by the restoration experiments and method comparison described in this work.