Image Denoising Methods Based On PDE And Graph-cut

Image denoising is a basic problem in image processing,the reaserch of PDE based denoising models and algorithms is an important part of it.So finding the stable and fast solving methods is theoretical significance and applied value.First his paper briefly introduced the Background knowledges such as the image denoising problem and existed models,additional operator splitting(AOS) method,graph method.Secondly,we introduce AOS scheme into the LLT(Lysaker,Lundervold,Tai) model, for improve the stability of the model.We proposed the AOS based LLT algorithm, obviously improve the stability,enlarge the selection range of time step.Accelerate the convergence speed under certain accuracy,achieve good result.Thirdly,we proposed a novel fourth-PDE based image denoising model based on the idea of LLT.The main idea is:we use different smoothing penalty coefficient to area with different smooth degree,design different smoothing functions for different purpose.And we also give the AOS based algorithm to solve it,improve the LLT model and have good stability.Finally,to reduce the stair-case of the gragh-cut based ROF model and void the e persecute.We introduce the gragh-cut method into LLT model,and design theα-(α+1) swap move space using the idea of move space,then establish the graph based move space algorithm which achieves good result.Besides this paper give the experiments which verify the validity and feasibility of above model and algorithm.