离散非周期傅里叶变换的思想就是将非周期信号拼接成为周期的离散信号来处理。如下图所示:
离散非周期信号
x
[
n
]
x[n]
x[n]拼接成周期信号
x
~
[
n
]
\widetilde{x}[n]
x
[n],
n
2
−
n
1
=
N
n2-n1=N
n2−n1=N。从而我们可以得到
x
~
[
n
]
\widetilde{x}[n]
x
[n]的傅立叶级数表示:
x
~
[
n
]
=
∑
k
=
<
N
>
X
k
e
i
k
2
π
N
n
\widetilde{x}[n] = \sum_{k=<N>}^{} X_{k}e^{ik\frac{2\pi}{N}n}
x
[n]=k=<N>∑XkeikN2πn
其中,
X
k
=
1
N
∑
n
=
<
N
>
x
~
[
n
]
e
−
i
k
2
π
N
n
X_{k} = \frac{1}{N} \sum_{n=<N>}^{}\widetilde{x}[n]e^{-ik\frac{2\pi}{N}n}
Xk=N1∑n=<N>x
[n]e−ikN2πn。由于
n
2
−
n
1
=
N
n2-n1=N
n2−n1=N,且
x
[
n
]
=
x
~
[
n
]
x[n] = \widetilde{x}[n]
x[n]=x
[n]在区间
[
n
1
,
n
2
]
[n1,n2]
[n1,n2],而对于区间
[
n
1
,
n
2
]
[n1,n2]
[n1,n2]之外的值,
x
[
n
]
=
0
x[n]=0
x[n]=0。所以得:
X
k
=
1
N
∑
n
=
n
1
n
2
x
[
n
]
e
−
i
k
2
π
N
n
=
1
N
∑
n
=
−
∞
∞
x
[
n
]
e
−
i
k
2
π
N
n
X_{k} = \frac{1}{N} \sum_{n=n1}^{n2}x[n]e^{-ik\frac{2\pi}{N}n} = \frac{1}{N} \sum_{n=-\infty}^{\infty}x[n]e^{-ik\frac{2\pi}{N}n}
Xk=N1n=n1∑n2x[n]e−ikN2πn=N1n=−∞∑∞x[n]e−ikN2πn
令
X
(
e
i
w
)
=
∑
n
=
−
∞
∞
x
[
n
]
e
−
i
w
n
X(e^{iw}) = \sum_{n=-\infty}^{\infty}x[n]e^{-iwn}
X(eiw)=∑n=−∞∞x[n]e−iwn,其中
w
=
k
w
0
=
k
2
π
N
w = kw_{0} = k\frac{2\pi}{N}
w=kw0=kN2π。得:
X
k
=
1
N
X
(
e
i
k
w
0
)
X_{k} = \frac{1}{N}X(e^{ikw_{0}})
Xk=N1X(eikw0)
将上式代入
x
~
[
n
]
\widetilde{x}[n]
x
[n]的傅立叶级数,得:
x
~
[
n
]
=
∑
k
=
<
N
>
1
N
X
(
e
i
k
w
0
)
e
i
k
2
π
N
n
=
1
2
π
∑
k
=
<
N
>
X
(
e
i
k
w
0
)
e
i
k
w
0
n
w
0
\begin{aligned} & \widetilde{x}[n] = \sum_{k=<N>}^{} \frac{1}{N}X(e^{ikw_{0}})e^{ik\frac{2\pi}{N}n} \\ & = \frac{1}{2\pi} \sum_{k=<N>}^{} X(e^{ikw_{0}})e^{ikw_{0}n}w_{0} \end{aligned}
x
[n]=k=<N>∑N1X(eikw0)eikN2πn=2π1k=<N>∑X(eikw0)eikw0nw0
另因为
k
k
k为整数,
d
k
dk
dk表示
k
k
k值一次的变化量,所以
d
k
=
1
dk=1
dk=1。那么
∑
k
=
<
N
>
…
d
k
\sum_{k=<N>}^{} \dots dk
∑k=<N>…dk等价于
∫
k
=
<
N
>
…
d
k
\int_{k=<N>}^{} \dots dk
∫k=<N>…dk。因此,上式继续推导:
x
~
[
n
]
=
1
2
π
∑
k
=
<
N
>
X
(
e
i
k
w
0
)
e
i
k
w
0
n
w
0
=
1
2
π
∫
k
=
<
N
>
X
(
e
i
k
w
0
)
e
i
k
w
0
n
w
0
d
k
=
1
2
π
∫
2
π
X
(
e
i
w
)
e
i
w
n
d
w
\begin{aligned} & \widetilde{x}[n] = \frac{1}{2\pi} \sum_{k=<N>}^{} X(e^{ikw_{0}})e^{ikw_{0}n}w_{0} \\ & = \frac{1}{2\pi} \int_{k=<N>}^{} X(e^{ikw_{0}})e^{ikw_{0}n}w_{0}dk \\ & = \frac{1}{2\pi} \int_{2\pi}^{} X(e^{iw})e^{iwn}dw \end{aligned}
x
[n]=2π1k=<N>∑X(eikw0)eikw0nw0=2π1∫k=<N>X(eikw0)eikw0nw0dk=2π1∫2πX(eiw)eiwndw
当
N
→
∞
N \rightarrow \infty
N→∞时,
x
~
[
n
]
=
x
[
n
]
\widetilde{x}[n] = x[n]
x
[n]=x[n]。至此,我们已经得到离散信号
x
[
n
]
x[n]
x[n]的傅里叶变换
X
(
e
i
w
)
X(e^{iw})
X(eiw),且信号
x
[
n
]
x[n]
x[n]为
X
(
e
i
w
)
X(e^{iw})
X(eiw)得逆变换。如下:
{
x
[
n
]
=
1
2
π
∫
2
π
X
(
e
i
w
)
e
i
w
n
d
w
X
(
e
i
w
)
=
∑
n
=
−
∞
∞
x
[
n
]
e
−
i
w
n
\begin{cases} & x[n] = \frac{1}{2\pi} \int_{2\pi}^{} X(e^{iw})e^{iwn}dw \\ & X(e^{iw}) = \sum_{n=-\infty}^{\infty}x[n]e^{-iwn} \end{cases}
{x[n]=2π1∫2πX(eiw)eiwndwX(eiw)=∑n=−∞∞x[n]e−iwn
2维离散非周期傅里叶变换
{ x [ m , n ] = 1 ( 2 π ) 2 ∫ 2 π ∫ 2 π X ( e i w 1 , e i w 2 ) e i ( w 1 m + w 2 n ) d w 1 d w 2 X ( e i w 1 , e i w 2 ) = ∑ m = − ∞ ∞ ∑ n = − ∞ ∞ x [ m , n ] e − i ( w 1 m + w 2 n ) \begin{cases} & x[m,n] = \frac{1}{(2\pi)^2} \int_{2\pi}^{}\int_{2\pi}^{} X(e^{iw_{1}},e^{iw_{2}})e^{i(w_{1}m+w_{2}n)}dw_{1}dw_{2} \\ & X(e^{iw_{1}},e^{iw_{2}}) = \sum_{m=-\infty}^{\infty} \sum_{n=-\infty}^{\infty}x[m,n]e^{-i(w_{1}m+w_{2}n)} \end{cases} {x[m,n]=(2π)21∫2π∫2πX(eiw1,eiw2)ei(w1m+w2n)dw1dw2X(eiw1,eiw2)=∑m=−∞∞∑n=−∞∞x[m,n]e−i(w1m+w2n)