PDE (Partial differential equation) is a widely used image denoising method. The anisotropic diffusion property of PDE is analyzed in this paper. For the different effects of forward and backward diffusion, some kinds of PDE smooth image well while some enhance edges. In this paper an adaptive image denoising model is proposed. The novel model diffuses differently based on gradient magnitude of image. The image parts which have large gradient magnitude are deemed as edges and textures, the adaptive model will enhance them. The selection of diffusion types changes gradually according to gradient magnitudes in different parts of image based on fuzzy theory. Both theoretical analysis and experiments have been used to verify that the novel adaptive image denoising model enhances edges while denoising.