Image Processing Reference
In-Depth Information
y
(
n
)
H
(
j
ω)
Q
g
(
n
)
f
(
n
)
FIGURE 2.35
Downsampling by a factor of Q.
2.7.4 S
AMPLING
R
ATE
C
ONVERSION BY A
F
ACTOR OF
Q
To change the sampling rate by a factor of
Q
,we
first upsample the signal by a factor
of P and then downsample it by a factor of Q, as shown in Figure 2.36.
The two LPFs can be combined into one
¼
P
and a bandwidth given by the minimum of
P
and
Q
. The overall system is shown in
Figure 2.37. In this
filter with a passband gain of P
1
figure, the LPF bandwidth is given by
P
,
Q
BW¼ min
(
2
:
116
)
2.7.5 E
XAMPLES OF
L
OW
-P
ASS
F
ILTERS
U
SED FOR
S
AMPLING
R
ATE
C
ONVERSION
An ideal LPF is not realizable, so in practice it is approximated by a simple realizable
FIR
filter. For example, the following simple
filters can be used for upsampling.
These
filters are derived using a linear interpolation algorithm. For example, for
upsampling by a factor of P, the LPF would be an FIR
filter of size 2P
1, given by
1
P
2
P
P
1
P
1
P
2
P
1
P
h
(
n
) ¼
1
(
2
:
117
)
P
The DC gain of this
filter is
X
X
DC gain ¼
H
(
j
v)j
v¼
0
¼
h
(
n
) exp (
j
v)j
v¼
0
¼
h
(
n
) ¼
P
(
:
)
2
118
n
n
g
(
n
)
H
1
(
j
ω)
H
2
(
j
ω)
Q
P
f
(
n
)
FIGURE 2.36
Sampling rate conversion by a factor of
Q
.
H
(ω)
P
g
(
n
)
Q
f
(
n
)
P
ω
-
BW
BW
FIGURE 2.37
Sampling rate conversion by factor of a
Q
.
Search WWH ::
Custom Search