An adaptation of widely-used image restoration methods has been presented, which uses a wavelet transform to suppress noise. The examples carried out on a range of images show the powerfulness of this approach. Unlike in various other approaches, there is no regularization coefficient to be estimated; and the problem of appropriate result is addressed directly in terms of noise and not in terms of smoothness or other such constraints. Photometry tests, using the approach described here, did not necessarily show an improvement over Richardson-Lucy restorations. Hence we propose, as work to be carried out in the future, an assessment of the usefulness of ``soft'' or adaptive alternatives to the noise suppression strategy adopted here, which was based on a ``hard'' threshold.