Study On Curvelet Tansform And Partial Differential Equations For Image De-noising

There are inevitably some noises in image because a lot of interferences exist in the process of getting and transmitting image. How to remove noises efficiently it has been a constant research topic in the fields of image processing. De-noising is important for improving image quality. Removing the noises and maintaining image characteristics is the core issue of image de-noising. Image de-noising models based on Curvelet transform and partial differential equations are studied deeply in this paper.The characteristic of Multiscale enable it to focus small changes of the image. Its direction parameter enble it to give expression to image edges and contours excellently. Those make Curvelet transform do a good work in the fields of image denosing.Partial differential equations for image de-noising base on the whole image. It can smooth the image under image features and their directions. So it can remove noises while preserve edges well. The P-M diffusion model and TV diffusion are introduced detailedly, they can keep the edge well but exist the shortcomings of staircasing effect, lose small details and textures.An image de-noising model by integrating anisotropic diffusion with Curvelet Transform is proposed in this paper by detailed analysis of Curvelet de-noising model and partial differential equations de-noising model. The method has combined the strongpoints of Curvelet Transform with partial differential equations de-noising method and conquered their disadvantages. It's main principal to improve the anisotropic diffusion de-noising model by means of dealing with the different matrixes of Curvelet Coefficient of the image by Curvelet transform via choosing appropriate Parameters of partial differential equations. It' s carried out the new anisotropic diffusion model based on the refined analysis of image. The result of the new model can avert the staircasing effect in the traditional anisotropic diffusion effectively and keep the textures and details of images better. This paper has argued that the new model has gotten higher PSNR and better visual quality than Curvelet de-noising and traditional anisotropic diffusion model.