Wavelet thresholding for nonnecessarily Gaussian noise: functionality

摘要

For signals belonging to balls in smoothness classes and noise with enoughmoments, the asymptotic behavior of the minimax quadratic risk amongsoft-threshold estimates is investigated. In turn, these results, combined witha median filtering method, lead to asymptotics for denoising heavy tails viawavelet thresholding. Some further comparisons of wavelet thresholding and ofkernel estimators are also briefly discussed.