Wavelet-based diffusion approaches for signal denoising

The efficiency of nonlinear diffusion filters is related to diffusivity. In the previous models, diffusivity depended on the derivatives of a signal, and hence it is easily affected by noise. This paper considers the nonlinear wavelet-based diffusion (NWD) methods for signal denoising in which the diffusivity is expressed by the wavelet transforms of received signal at several scales. Due to the multiresolution of wavelet transform, the proposed diffusivity with wavelet transforms efficiently reduces the influence of noise on the estimate of diffusion amount and improves nonlinear diffusion filters. Some numerical experimental results compared with the previous models are shown.