Wavelet-based signal de-noising via simple singularities approximation

In this paper WISDOW (wavelet-based image and signal de-noising via overlapping waves) is presented. It consists of a novel model for noise removal using wavelets. Its main peculiarity is the modelling of a noisy signal as composition of elementary atoms which behave as interfering waves in the wavelet domain. Signal recovery is then performed by means of the overlapping effects principle at each scale level. Early theoretical and experimental results show the great potential of the proposed model.