denoise: Wavelet-based Denoising

In rwt: 'Rice Wavelet Toolbox' Wrapper

Wavelet-based Denoising

Description

Denoise the signal x using the 2-band wavelet system described by the filter h using either the traditional discrete wavelet transform (DWT) or the linear shift invariant discrete wavelet transform (also known as the undecimated DWT (UDWT)).

Usage

denoise(x, h, type, option)
denoise.dwt(x, h, option = default.dwt.option)
denoise.udwt(x, h, option = default.udwt.option)

Arguments

x	1D or 2D signal to be denoised
h	numeric scalar specifying scaling filter to be applied
type	type of transform. Valid values are: DWT.TRANSFORM.TYPE UDWT.TRANSFORM.TYPE
option	list containing desired transformation settings

Details

The transformation settings in the option list are:

threshold.low.pass.part:

logical scalar. If TRUE, threshold the low-pass component.

threshold.multiplier:

thld = c*MAD(noise_estimate)

variance.estimator:

Valid values are:

MAD.VARIANCE.ESTIMATOR	Mean absolute deviation
STD.VARIANCE.ESTIMATOR	Classical numerical std estimate

threshold.type:

Valid values are:

SOFT.THRESHOLD.TYPE	Soft thresholding
HARD.THRESHOLD.TYPE	Hard thresholding

num.decompression.levels:

number of levels in wavelet decomposition. Setting this to MAX.DECOMPOSITION will allow maximal decomposition.

threshold:

actual threshold to use. Setting this to anything but CALC.THRESHOLD.TO.USE will disable the variance.estimator setting.

Value

Returns a list with components:

xd	estimate of noise free signal
xn	estimated noise signal (x-xd)
option	list of actual parameters used. It is configured the same way as the input option list with an additional element - option[[7]] = type.

Note

Both denoise.dwt and denoise.udwt are convenience routines that call the denoise routine with appropriate default arguments.

Author(s)

P. Roebuck [e-mail]

Examples

sig <- makesig(SIGNAL.DOPPLER)
h <- daubcqf(6)
ret.dwt <- denoise.dwt(sig$x, h$h.0)

rwt documentation built on June 14, 2022, 5:07 p.m.