Two Image Denoising Method Of Based On Partial Ifferential Equations

At present, partial differential equations has become an important method in the field ofimage analysis. Denoising is a very important part of the image processing is of greatsignificance for improving the visual effects and image quality of the image. Partialdifferential equations for image denoising method to carry out research work, the imagedenoising method based on partial differential equations of the PM model and the Gaussiancurvature-driven diffusion model.1.Image denoising is a very important research topic in the research field of imageprocessing. Analysis of the traditional image denoising method based on the traditionalmethod for image denoising while also destroy the edge detail and texture characteristics ofthe image information. In this paper, the P-M model for how to select the appropriatetheoretical basis of the diffusion coefficient; Based on the Gaussian curvature drivendiffusion model of the conduction depends only on the Gaussian curvature value, withouttaking into account the influence of the gradient, so the model will inevitably lead to anonzero the value of Gaussian curvature gradient edge blur, while the rate of diffusion cannot be well controlled in the region of the image different gradient values. Denoising modelin accordance with its shortcomings, the introduction of Tukey’s biweight function andadaptive diffusion coefficient of the TV model to control the rate of diffusion, the newGaussian curvature driven model.2.P-M model and the improved Gaussian curvature-driven diffusion model based on theblend model fusion model in addition to being able to play the advantages of the PM modelin denoising the same time to maintain the edge, while the Gaussian curvaturedrivenGaussian curvature of the smooth region of the model the details of the characteristicsof the protective effect of the continuity of the model combines the advantages of both, at thesame time better able to overcome the PM model, the threshold in the denoising process,select the larger will cause the image fuzzy and Gaussian curvature-driven diffusion modelin the denoising process the denoising shortcomings of black and white dots. Theexperimental results show that the denoising can also be good to remove the black and whitepoints, improving the signal to noise ratio of the image processing, fusion model of differentgradient area the size of the corresponding processing, this not only to overcome the use ofsalt and pepper noise can not be removed by simple curve fitting method, and can betterkeep the details of the image features, the fusion model has better denoising effect.