accessD.mwd: Get wavelet coefficients from multiple wavelet structure...

In wavethresh: Wavelets Statistics and Transforms

Get wavelet coefficients from multiple wavelet structure (mwd).

Description

The wavelet coefficients from a multiple wavelet decomposition structure, mwd.object, (e.g. returned from mwd) are packed into a single matrix in that structure. This function extracts the coefficients corresponding to a particular resolution level.

Usage

## S3 method for class 'mwd'
accessD(mwd, level, ...)

Arguments

mwd	Multiple wavelet decomposition structure from which you wish to extract the expansion coefficients.
level	The level that you wish to extract. If the "original" data has mwd$filter$npsi*2^m data points (mwd$filter$npsi being the multiplicity of the multiple wavelets) then there are m possible levels that you could want to access, indexed by 0,1,...,(m-1)
...	any other arguments

Details

The mwd function produces a multiple wavelet decomposition object .

The need for this function is a consequence of the pyramidal structure of Mallats algorithm and the memory efficiency gain achieved by storing the pyramid as a linear matrix. AccessD obtains information about where the coefficients appear from the fl.dbase component of mwd, in particular the array fl.dbase$first.last.d which gives a complete specification of index numbers and offsets for mwd$D.

Note that this function and accessC only work on objects of class mwd to extract coefficients. You have to use putD.mwd to insert wavelet coefficients into a mwd object.

See Downie and Silverman, 1998.

Value

A matrix with mwd$filter$npsi rows containing the extracted coefficients.

RELEASE

Tim Downie 1995-6

Author(s)

G P Nason

See Also

accessD.mwd, draw.mwd, mfirst.last, mfilter.select, mwd, mwd.object, plot.mwd, print.mwd, putC.mwd, putD.mwd, summary.mwd, threshold.mwd, wd, wr.mwd

Examples

#
# Get the 3rd level of smoothed data from a decomposition
#
data(ipd)
accessD.mwd(mwd(ipd), level=3)

wavethresh documentation built on Sept. 11, 2024, 9:33 p.m.