Noise Removal With Anisotropic Diffusion

It is a basic and important work in the digital image processing engineering to recover the original image as well as possible from the noised observation image. Many engineering applications of digital image have a great deal to do with denoising an image. Noise removal has taken a lot of attention of related researchers.Classical image denoising methods not only remove the noise of an image but also blur some it's edges and textures. This is not what we want. We hope to get a method that not only is able to remove the noise but also can keep edges and textures.In recent years, some researchers remove noise of an image with partial differen-tial equations, have greatly developed this field and achieved many good results.One of these successful methods is diffusion.In this paper, to remove noise from a noisy image, we introduce a new anisotropic diffusion method based on Euler's elastica functional which has been used when we inpaint an image. We have tested the method on a broad range of images. These numerical examples indicate that this method can remove almost all noise and keep more edges and textures.