Some regularization Strategies for an Ill-Posed Denoising Problem
Khuda Bux Amur · International journal of tomography and simulation · 2012
A nonlinear problem is considered for image denoising which is ill-posed. We propose a novel adaptive strategy for local adaptive choice of optimal regularization parameters to identify the damaged regions of the image domain where the average L^2 error is large or greater than a given tolerance during the image denoising process. This regularization strategy decreases the diffusion error and improves the visual quality of image at each adaptive iteration.The numerical scheme for this method is based on adaptive finite element method and unstructured image grid as discrete domain of computation. In the first case the regularization parameters are manually fixed where the diffusion is tested for various time iterations. In the second case we propose an intelligent adaptive algorithm which automatically identify the damaged regions of the image domain and performs the regularization in these regions.