This paper puts forward a new edge-directed enhancing based anisotropic diffusion model.Since edge enhancing diffusion can not get right edge direction

and coherence enhancing diffusion often induces false edges in image denoising

our new model adapts a new operator for edge orientation based on nonlinear smoothing operator

and defines eigenvalues on edge's orientation such that the new tensor has large eigenvalues in both gradient direction and edge direction on the slippy region of image

but has small eigenvalues in gradient direction and large eigenvalues in edge direction on the edge region of image.So it can remove noise efficiently and keep edges better at the same time to get good result on the whole image.Both theory analysis and numerical results show that the new model has better denosing results than the known diffusion model

and has higher peak signal to noise ratio and higher edge preserved index.