tguh.denoise {breakfast}R Documentation

Noise removal from Tail-Greedy Unbalanced Haar coefficients via connected thresholding

Description

This function performs the connected thresholding of the Tail-Greedy Unbalanced Haar coefficients.

Usage

tguh.denoise(tguh.decomp.obj, lambda, minseglen = 1, bal = 1/20)

Arguments

tguh.decomp.obj

A variable returned by tguh.decomp or tguh.denoise.

lambda

The threshold value.

minseglen

The minimum permitted length of either wing of any Unbalanced Haar wavelet whose corresponding coefficient survives the thresholding.

bal

The minimum permitted ratio of the length of either wing to the sum of the lengths of both wings of any Unbalanced Haar wavelet whose corresponding coefficient survives the thresholding.

Details

Typically, the first parameter of tguh.denoise will be an object returned by tguh.decomp. The function tguh.denoise performs the "connected thresholding" of this object, in the sense that if a Tail-Greedy Unbalanced Haar detail coefficient does not have any surviving children coefficients, then it gets set to zero if it falls under the threshold, or if the corresponding Unbalanced Haar wavelet is too unbalanced or has too short a wing. See "Tail-greedy bottom-up data decompositions and fast multiple change-point detection", P. Fryzlewicz (2017), preprint, for details.

Value

Modified object tguh.decomp.obj; the modification is that the detail coefficients in the decomp.hist field that do not survive the thresholding get set to zero.

Author(s)

Piotr Fryzlewicz, p.fryzlewicz@lse.ac.uk

See Also

tguh.cpt, tguh.decomp, tguh.reconstr

Examples

rnoise <- rnorm(10)
rnoise.tguh <- tguh.decomp(rnoise)
print(rnoise.tguh)
rnoise.denoise <- tguh.denoise(rnoise.tguh, 3)
rnoise.clean <- tguh.reconstr(rnoise.denoise)
print(rnoise.clean)

[Package breakfast version 1.0.0 Index]