作者: Alessandro foi , Radu Bilcu , Vladimir Katkovnik , Karen Egiazarian
DOI: 10.1109/NORSIG.2006.275285
关键词:
摘要: We presenet a new transform-based methdo for adaptive denoising. It is assumed that the observatioons are given by broad class of models with signal-dependent variance. Denoising performed coefficient shrinkage in local block-transform domain. The intersection confidence intervals (ICI) rule used order to determine spatially-adaptive size block transforms. enables both simpler modelling noise transform domain and sparser decomposition signal. Consequently, very effective reconstructed estimate's quality high. Experiments simulated as well real data demonstrate advanced performance proposed algorithm.