Digital Signal Processing Reference
In-Depth Information
Ψ
(
ω
)
ψ
(t/a)
0 < a < 1
0 < a < 1
t
ω
Ψ
( )
ω
ψ
(t/a)
a > 1
a > 1
t
ω
(b)
(a)
Figure 2.2
Wavelet in time and frequency domains: (a) scale parameter 0
<a<
1, (b) scale
parameter
a>
1.
frequency resolution of the WT can easily be detected.
The foundation of the WT is based on the scaling property of the
Fourier transform. If
ψ
(
t
)
←→
Ψ(
ω
)
represents a Fourier transform pair, then we have
√
a
Ψ
t
←→
√
a
Ψ(
aω
)
1
(2.22)
a
with
a>
0 being a continuous variable. A contraction in the time domain
produces an expansion in the frequency domain, and vice versa. Figure
2.3 illustrates the corresponding resolution cells in the time-frequency
domain. The figure makes visual the underlying property of wavelets:
they are localized in both time and frequency. The functions
e
jωt
are
Search WWH ::
Custom Search