In this paper, we present a multiresolution approach for color image de-noising by image reconstruction. The proposed method relies on a modified wavelet reconstruction algorithm. This algorithm does not use the wavelet maxima generated by decomposition but a new maxima representation. This new representation is calculated from a multi-scale region segmentation approach. The multi-scale segmentation process allows us to obtain closed edges (vital for a correct image reconstruction) and a more simple selection of most significant edges (the less affected by noise). The image de-noising result is judged by the uniformity of the reconstructed image regions.
Adrian Stoica, Philippe Carré, Christine Fernandez-Maloigne, "Color Images De-Noising by Wavelet Maxima Representation and Regions Segmentation" in Proc. IS&T CGIV 2002 First European Conf. on Colour in Graphics, Imaging, and Vision, 2002, pp 375 - 379, https://doi.org/10.2352/CGIV.2002.1.1.art00079