Geoscience Reference
In-Depth Information
Figure 6.2 The filter function, Eq. (6.12) , and transfer function, Eq. (6.11) ,ofa
one-dimensional, low-pass, wave-cutoff filter.
T and G for a one-dimensional, low-pass wave-cutoff filter are plotted in
Figure 6.2 .
From Eq. (6.6) low-pass wave-cutoff filtering gives in physical space
x )
sin
[
κ c (x
]
f r (x)
x )f (x )dx =
f(x )dx .
=
G(x
(6.13)
x )
π(x
−∞
−∞
Figure 6.2 shows that this filter function G is negative in some regions. This
is required in order that
the filter have a sharp cutoff in the wavenumber
domain.
The high-pass wave-cutoff filter rejects Fourier components of smaller wave-
number and passes those of larger wavenumbers:
T(κ)
=
0 ,
κ
κ c ;
T(κ)
=
1 ,
κ> c .
(6.14)
6.2.4 The Gaussian filter
Both the running-mean filter, Figure 6.1 , and the wave-cutoff filter, Figure 6.2 , have
negative values in physical or wavenumber space. One that does not is the Gaussian
filter,
1
x 2
2 σ 2 .
2 πσ e
G(x)
=
(6.15)
Its transfer function is also Gaussian (Problem 6.25) .
 
Search WWH ::




Custom Search