In both digital filter design and spectral estimation, the choice of a windowing function can play an important role in determining the quality of overall results. The main role of the window is to damp out the effects of the Gibbs phenomenon that resulted from truncation of an infinite series.
| Function | Window Type |
|---|---|
| barthann | Modified Bartlett-Hann |
| bartlett | Bartlett |
| blackman | Blackman |
| blackmanharris | Minimum four-term Blackman-Harris |
| bohman | Bohman |
| Chebyshev | |
| flattop | Flat top weighted |
| hamming | Hamming |
| hanning | Hann (Hanning) |
| nuttall | Nuttall-defined minimum 4-term Blackman-Harris |
| parzen | Parzen (de la Vallée Poussin) |
| rectangular | Rectangular |
| triangular | Triangular |
| tukey | Tukey (tapered cosine) |
| Taylor | |
| gauss | Gaussian |
| kaiser | Kaiser |
| Equivalent noise bandwidth |
To install the released stable version, run
using Pkg
Pkg.add("WindowFunctions")using Plots
using WindowFunctions
kaiser5(N; dtype::DataType=Float32) = kaiser(N, 5.0, 10, dtype=dtype)
winlist = [barthann,
bartlett,
blackman,
blackmanharris,
bohman,
flattop,
hamming,
hanning,
nuttall,
parzen,
rectangular,
triangular,
gauss,
tukey,
kaiser5]
for win in winlist
plot(win(256, dtype=Float32),
label=string(win),
fillrange = 0,
fillalpha = 0.3,
show=true)
ylims!(-0.13,1.18)
xlims!(1-20,256+20)
sleep(1)
end