Image Denoising and Segmentation using COMSOL Multiphysics

F. Zama
Department of Mathematics, Bologna University, Bologna, Italy

Partial differential equations have recently become popular and useful tools for several image processing tasks such as image de-noising and segmentation.

In this work, we implement a unified image de-noising and segmentation approach which is based on a nonlinear diffusion equation with a reactive term for achieving edge preserving smoothing and segmentation.

This model is highly nonlinear and the computation of the Gaussian low pass filter at the intermediate time steps is computationally very time consuming.

In order to speed up the process, a coupled system is efficiently implemented using COMSOL Multiphysics. The filter obtained has a variance parameter changing in the integration interval.