See Nason, Sapatinas and Sawczenko, 1998 for further details on ordering and weaving.
Note that the output object will be of the non-decimated type. In other
words the type component of the output object will be
"station".
Once the input object has been converted the output can be used
with any of the functions suitable for the wd object
with type component equal to "station".
The actual weaving permutation for shuffling coefficients from one representation to another is achieved by the getarrvec function.
# # Generate a sequence of 32 random normals (say) and take their # packet-ordered non-decimated wavelet transform # myrand <- wst(rnorm(32)) # # Print out the result (to verify the class and type of the object) # myrand #Class 'wst' : Stationary Wavelet Transform Object: # ~~~ : List with 5 components with names # wp Carray nlevels filter date # #$wp and $Carray are the coefficient matrices # #Created on : Tue Sep 29 12:29:45 1998 # #summary(.): #---------- #Levels: 5 #Length of original: 32 #Filter was: Daub cmpct on least asymm N=10 #Date: Tue Sep 29 12:29:45 1998 # # Yep, the myrand object is of class: wst object. # # Now let's convert it to class wd. The object # gets returned and, as usual in S, is printed. # convert(myrand) #Class 'wd' : Discrete Wavelet Transform Object: # ~~ : List with 8 components with names # C D nlevels fl.dbase filter type bc date # #$ C and $ D are LONG coefficient vectors ! # #Created on : Tue Sep 29 12:29:45 1998 #Type of decomposition: station # #summary(.): #---------- #Levels: 5 #Length of original: 32 #Filter was: Daub cmpct on least asymm N=10 #Boundary handling: periodic #Transform type: station #Date: Tue Sep 29 12:29:45 1998 # # The returned object is of class wd with a # type of "station". # I.e. it has been converted successfully.