On the Choice of Thresholding Parameters for Non-Gaussian Noise Distribution

被引:0
|
作者
Kudryavtsev A.A. [1 ]
Shestakov O.V. [1 ,2 ]
机构
[1] Department of Mathematical Statistics, Faculty of Computational Mathematics and Cybernetics, M. V. Lomonosov Moscow State University, Moscow
[2] Institute of Informatics Problems, Federal Research Center “Computer Science and Control” of the Russian Academy of Sciences, Moscow
关键词
D O I
10.1007/s10958-020-04756-7
中图分类号
学科分类号
摘要
The paper considers the problem of estimating the signal function from noisy observations using threshold processing of its wavelet expansion coefficients. Under general assumptions about the properties of the noise distribution, the asymptotic order of the optimal threshold is calculated, minimizing the loss function, based on the probability that the maximum error in the wavelet coefficients exceeds a given critical level. © 2020, Springer Science+Business Media, LLC, part of Springer Nature.
引用
收藏
页码:519 / 524
页数:5
相关论文
共 50 条