dt2cen: centre of the DT-spectrum

In dualtrees: Decimated and Undecimated 2D Complex Dual-Tree Wavelet Transform

centre of the DT-spectrum

Description

calculate the centre of mass of the local spectra in hexagonal geometry

Usage

dt2cen(dt, mask = NULL)

Arguments

dt	a J x nx x ny x 6 array of spectral energies, the output of fld2dt
mask	a nx x ny array of logical values

Details

Each of the J x 6 spectral values is assigned a coordinate in 3D space with x(d,j)=cos(60*(d-1)), y(d,j)=sin(60*(d-1)), z(d,j)=j, where j denotes the scale and d the direction. Then the centre of mass in this space is calculated, the spectral values being the masses at each vertex. The x- and y-cooridnate are then transformed into a radius rho=sqrt(x^2+y^2) and an angle phi=15+0.5*atan2(y,x). rho measures the degree of anisotropy at each pixel, phi the orientation of edges in the image, and the third coordinate, z, the central scale. If a mask is provided, values where mask==TRUE are set to NA.

Value

a nx x ny x 3 array where the third dimension denotes degree of anisotropy, angle and central scale, respectively.

Note

Since the centre of mass is not defined for negative mass, any values below zero are removed at this point.

Examples

dt <- fld2dt(blossom)
ce <- dt2cen(dt)
image( ce[,,3], col=gray.colors(32, 0, 1) )

dualtrees documentation built on Oct. 4, 2023, 5:10 p.m.