stheoreme (version 1.2)

d2spec: Power Spectrum Probability Vector Calculator for Matrices

Description

Function d2spec is applied to a pair of matrices (2d arrays) and generates the pair of corresponding 1d spectral power density functions (to be used then as probability vectors) specified by the method argument

Usage

d2spec(d2arr0, d2arr1, band=c(0,0), brks=0, method='rad', meansub=TRUE)

Arguments

d2arr0

sample matrix

d2arr1

sample matrix

band

two border values to set a range of considered frequencies. The default c(0,0) sets full entire range from 0 to 0.5 Hz (where 1 Hz = [1/sampling_interval]) for radial method and from 0 to 180 degrees for both angular methods

brks

value used in a same manner as the number of cells for the histogram. The default brks=0 sets the number of cells equal to min(length(sample0), length(sample1))/2 for radial method and equal to 180 for both angular methods

method

specifies the method of using the power spectrum as a probability vector. The default value 'rad' defines the calculation of spectral power density for spatial frequency space {u,v} as a function of spectral radius sqrt(u^2+v^2).

meansub

logical item defining if individual baselines (as mean value) are subtracted from original matrices before application of 2d Fourier transform

Value

• f0
  spectral power density function as a probability vector representing state0 of a system
• f1
  spectral power density function as a probability vector representing state1 of a system
• efs
  vector of corresponding metrics/bin values (radial frequency or spectral angle)

Details

Spectral power density is often being used as probability vector/matrix characterizing a state of a system, e.g. in image analysis. Here 2d fast Fourier transform algorithm is utilized to calculate power spectrum matrix $P(u,v)$ and then the 1d spectral power density: $$f(k_{i})=\sum_{k_{min}}^{k_{max}}{ P(k: k_{i}method='rad'); $k = atan(u/v)$ (as angle where method='ang') or $k = atan(v/u)$ (as angle where method='ang90')

and where ${k_{min},k_{max}}$ are defined by band argument

In general the function works similarly to d1nat. As a bonus it prints basic statistics summary for spectral power density functions alongside with technical plot. It is recommended for use as a data preparation step before following Klimontovich's S-theorem based analysis.

References

N.F.Zhang, A.E.Vladar, M.T.Postek, R.D.Larrabee.Spectral density-based statistical measures for image sharpness. Metrologia, 42(2005), 351-359

E.J.Groth. Probability distributions related to power spectra. Astrophysical Journal, 1975. Suppl. Ser., Vol.29, No.286, p. 285-302.

See Also

crit.stheorem, cxds.stheorem, d1spec, d1nat

Examples

# compare (harmonic+background) with (harmonic noisy period) matrices
s0<-array(2+sin(c(1:256)/3), c(16,16))
s1<-array(sin(c(1:512)+runif(512,0,2)), c(16,32))

# as radial:
b<-d2spec(d2arr0=s0, d2arr1=s1); b
b<-d2spec(s0, s1, brks=29, band=c(0.15,0.5)); b
# as angular:
b<-d2spec(s0, s1, method='ang', meansub=FALSE); b

#example of 3-step data analysis with Klimontovich's S-theorem
# Study two gratings: random vs regular
s0<-array(c(rep(0,640),rep(1,640)), c(320,320))
s1<-array(runif(5120,0,1), c(64,80))
# step a. Binarize (to make s1 comparable with s0 by its nature as a grating)
a<-utild2bin(s0, s1, method='med')
# step b. Create probability vectors as for angular space (anisotropy study)
# There is no doubt s0 is more regular
b<-d2spec(a$bin0, a$bin1, brks=36, method='ang90'); b
# step c. Compare gratings with Klimontovich's S-theorem. Renormalized entropy shift
# is negligible compared to Shannon's. Evolution from state0 to state1 is possible
# but clearly with external entropy (or energy) inflow
crit.stheorem(b$f0,b$f1)
cxds.stheorem(b$f0,b$f1)