Image De-noising Algorithms Based on PDE and Wavelet

The traditional PDE based de-nosing models detected edges by the gradients of images, and they were easily affected by noise. Combining PDE with wavelet, we developed three de-noising schemes for images. In the first proposed model, a diffusion function was introduced in the regularization term of the ROF model, and the modulus of gradient was substituted by the modulus of wavelet transform, which gave results that the new model could preserve edges better and had strong ability of resisting noise. But this new model required high computational effort, considered the features of noise in wavelet domain, we proposed the second models to reduce computational complexity. The last new model was presented based on the character of the multi-resolution analysis of wavelet transform. The experimental results show improvements of all the proposed models.