Abstract: Based on the standard ROF model, an improved method for image de-noising is proposed. In order to overcome the weakness that edge information is sensitive to noise and prone to blur of the ROF model while de-noising, in the new algorithm, the anisotropic diffusion function is introduced, and the magnitude of wavelet transform is substituted for the module of the gradient operator, which makes the new model more robust and diffuse according to the characteristic of the image, and therefore, the edges are preserved better. Experiments show that there is an increase of about 1.5dB on average in the signal to noise ratio and a remarkable improvement on the visual effect.

Key words: ROF model, image de-noising, wavelet diffusion