imwdc.object {wavethresh} | R Documentation |
These are objects of classes
imwdc
They represent a decomposition of an image with respect to a two-dimensional wavelet basis
In previous releases the original image was stored as the "original" component of a imwd object. This is not done now as the resulting objects were excessively large.
To uncompress this class of object back into an object of class imwd.object
use the uncompress.imwdc
function.
The following components must be included in a legitimate `imwdc' object.
nlevels |
number of levels in wavelet decomposition. If you raise 2 to the power of nlevels then you get the dimension of the image that you originally started with. |
type |
If type="wavelet" then the image was decomposed according to the 2D Mallat pyramidal algorithm. If type="station" then the image was decomposed using the 2D spatially ordered non-decimated wavelet transform. |
fl.dbase |
The first last database associated with the decomposition. For images, this list is not very useful as each level's components is stored as a list component, rather than being packaged up in a single vector as in the 1D case. Nevertheless the internals still need to know about fl.dbase to get the computations correct. See the help for first.last if you are a masochist.
filter.select function. This component records the filter used in the decomposition. The reconstruction routines use this component to find out what filter to use in reconstruction. Each vector stores its coefficients using an object of class compressed, i.e. the vector is run-length encoded on zeroes. Note that the levels should be in numerically decreasing order, so if nlevels is 5, then there will be w5L1, w5L2, w5L3 first, then down to w1L1, w1L2, and w1L3. Note that these coefficients store their data according to the first.last database "fl.dbase$first.last.d", so refer to them using this.
Note that if type="wavelet" 2^N pixels. If the type component is "station" then each coefficient subimage is of the same dimension as the input image used to create this object. |
w0Lconstant |
This is the coefficient of the bottom level scaling function coefficient. So for example, if you used Haar wavelets this would be the sample mean of the data (scaled by some factor depending on the number of levels, nlevels). |
bc |
This component details how the boundaries were treated in the decomposition. |
This class of objects is returned from the threshold.imwd
function to represent a thresholded two-dimensional wavelet decomposition of a function. Some other functions return an object of class imwdc.
The imwd class of objects has methods for the following generic functions: draw
, imwr
, nullevels
, plot
, print
, summary
, threshold.imwdc
.
Version 3.5.3 Copyright Guy Nason 1994
imwd
imwd.object
, threshold.imwd
, uncompress.imwdc
.