ESTIMATING LAW OF VARIATION OF HETEROSCEDASTIC NOISE IN TRAJECTORY MEASUREMENTS ON THE BASIS OF WAVELETS
In this paper we describe a two-step method of denoising of comlex data in heteroscedastic nonparametric regression. At the first step we extract using wavelets a noise component and estimate for its absolute values using robust spline fitting the model for the variance function. At the second step we employ the estimated variance function for wavelet coefficients thresholding on the basis of overlap discrete wavelet transform.
Authors: N. I. Oreshko
Direction: Informatics, management and Computer Technology
Keywords: Denoising, heteroscedastic noise, wavelet transform
View full article