dualtrees (version 0.1.5)

fld2dt: transform a field into an array of spectral energies

Description

Handles the transformation itself, boundary conditions and bias correction and returns the unbiased local wavelet spectrum at each grid-point.

Usage

fld2dt(fld, Nx = NULL, Ny = NULL, J = NULL, correct = TRUE,
  rsm = 0, verbose = FALSE, boundaries = "pad",
  fb1 = near_sym_b_bp, fb2 = qshift_b_bp)

Value

an array of size J x nx x ny x 6 where dim(fld)=c(nx,ny)

Arguments

fld

a real matrix

Nx

size to which the field is padded in x-direction

Ny

size to which the field is padded in y-direction

J

number of levels for the decomposition

correct

logical, whether or not to apply the bias correction

rsm

number of pixels to be linearly smoothed along each edge before applying the boundary conditions (see smooth_borders).

verbose

whether or not you want the transform to talk to you

boundaries

how to handle the boundary conditions, either "pad", "mirror" or "periodic"

fb1

filter bank for level 1

fb2

filter bank for all further levels

Details

The input is blown up to Nx x Ny and transformed by dtcwt(..., dec=FALSE). Then the original domain is cut out, the coefficients are squared and the bias is corrected (for details on the bias, see A).

References

Nelson, J. D. B., A. J. Gibberd, C. Nafornita, and N. Kingsbury (2018) <doi:10.1007/s11222-017-9784-0>

See Also

A

Examples

oldpar <- par( no.readonly=TRUE )
dt <- fld2dt( blossom )
par( mfrow=c(2,2), mar=rep(2,4) )
for( j in 1:4 ){
    image( blossom, col=gray.colors(128, 0,1), xaxt="n", yaxt="n" )
    for(d in  1:6) contour( dt[j,,,d], levels=quantile(dt[,,,], .995), 
                            col=d+1, add=TRUE, lwd=2, drawlabels=FALSE )
    title( main=paste0("j=",j) )
} 
x0  <- seq( .1,.5,,6 )
y0  <- rep( 0.01,6 )
a   <- .075
phi <- seq( 15,,30,6 )*pi/180
x1  <- x0 + a*cos( phi )
y1  <- y0 + a*sin( phi )
rect( min(x0,x1)-.05, min(y0,y1)-.05, 
      max(x0,x1)+.05, max(y0,y1), col="black", border=NA )
arrows( x0, y0, x1, y1, length=.05, col=2:7, lwd=2, code=3 )
par( oldpar )