Fourier Transform

Fourier Transform

基本的定义

严格来说, 傅里叶变换是Schwartz space 上的一一映射, 对于$L^1$, 即可积函数我们都可以找到其对应的傅里叶变换.

符号	定义
傅里叶变换: $\hat{f}(u)$	${\int }_{-\infty }^{+\infty }f(t)e^{-j2\pi ut}\mathrm{d}t$
逆傅里叶变换: $\check{F}(t)$	${\int }_{-\infty }^{+\infty }F(u)e^{j2\pi tu}\mathrm{d}u$​
离散傅里叶变换: $F_m={\hat{f}}_m$	${\sum }_{n=0}^{N-1}{f}_n{e}^{-j2\pi nm/N}$​
离散逆傅里叶变换: $f_n={\check{F}}_n$	$\frac{1}{N}{\sum }_{m=0}^{N-1}{F}_m{e}^{j2\pi mn/N}$
卷积: $f\star g(t)$	${\int }_{-\infty }^{+\infty }f(\tau )g(t-\tau )\mathrm{d}\tau$
correlation: $f\bullet g(t)$​	${\int }_{-\infty }^{+\infty }{f}^{\ast }(\tau )g(\tau +t)\mathrm{d}\tau$
二元傅里叶变换: $\hat{f}(u,v)$	${\int }_{-\infty }^{+\infty }{\int }_{-\infty }^{+\infty }f(t,z)e^{-j2\pi ut}{e}^{-j2\pi vz}\mathrm{d}t\mathrm{d}z$
二元傅里叶逆变换: $\check{F}(t,z)$	${\int }_{-\infty }^{+\infty }{\int }_{-\infty }^{+\infty }F(u,v)e^{j2\pi tu}{e}^{-j2\pi zv}\mathrm{d}t\mathrm{d}z$
二元卷积: $f\star g(t,z)$	${\int }_{-\infty }^{+\infty }{\int }_{-\infty }^{+\infty }f(\tau ,\omega )g(t-\tau ,z-\omega )\mathrm{d}\tau \mathrm{d}\omega$
共轭	$f^{\ast }(t)=[r(t)+ji(t){]}^{\ast }=r(t)-ji(t),r(t),i(t)\in \mathbb{R}$

注: 不同版本的傅里叶变换的定义存在系数上的差异.

性质

	$\hat{}$	$\stackrel{\mathcal{F}}{\iff }$	$\stackrel{ˇ}{}$
1	$f(t)$	$\stackrel{\mathcal{F}}{\iff }$	$F(u)$
2	$f\star g(t)$	$\stackrel{\mathcal{F}}{\iff }$	$F(u)\cdot G(u)$
3	$f\bullet g(t)$​	$\stackrel{\mathcal{F}}{\iff }$	$F^{\ast }(u)\cdot G(u)$
4	$f(t) \cdot g(t) $	$\stackrel{\mathcal{F}}{\iff }$	$F\star G(u)$
5	$f(t)+g(t)$	$\stackrel{\mathcal{F}}{\iff }$	$F(u)+G(u)$
6	$f(t+t_0)$	$\stackrel{\mathcal{F}}{\iff }$	$e^{j2\pi t_{0}u}F(u)$
7	$f(-t)$	$\stackrel{\mathcal{F}}{\iff }$	$F(-u)$
8	$F(u)=\hat{f}(u)$​	$\stackrel{\mathcal{F}}{\iff }$	$f(-t)=\hat{\hat{f}}(t)$
以上	对于	离散	成立
9	$f(at)$	$\stackrel{\mathcal{F}}{\iff }$	$\frac{1}{
10	$f^{\prime }(t)$	$\stackrel{\mathcal{F}}{\iff }$	$j2\pi uF(u)$
11	$-j2\pi tf(t)$	$\stackrel{\mathcal{F}}{\iff }$	$F^{\prime }(u)$
12	$f(R\mathbf{t})$, $R^{T}R=I$	$\stackrel{\mathcal{F}}{\iff }$	$F(R\mathbf{u})$

注: 虽然以上性质除11外都是用了标量, 但是实际上对于一般的傅里叶变换也是成立的.

1. pass

2. 卷积:
   $$\begin{array}{ll}(f\star g{)}^{\hat{}}(u)& =\int f\star g(t)e^{-j2\pi ut}\mathrm{d}t\\ & =\int \int f(t-\tau )g(\tau )\mathrm{d}\tau e^{-j2\pi ut}\mathrm{d}t\\ & =\int \int f(t-\tau )g(\tau )e^{-j2\pi ut}\mathrm{d}t\mathrm{d}\tau \to \mathrm{F}\mathrm{u}\mathrm{b}\mathrm{i}\mathrm{n}\mathrm{i}\\ & =\int \int f(t-\tau )e^{-j2\pi u(t-\tau )}\mathrm{d}tg(\tau )e^{-j2\pi u\tau }\mathrm{d}\tau \\ & =F(u)\cdot G(u).\end{array}$$
   $\Leftarrow$是类似的证明.

3. correlation:
   $$\begin{array}{ll}(f\bullet g{)}^{\hat{}}(u)& =\int f\ast g(t)e^{-j2\pi ut}\mathrm{d}t\\ & =\int \int f^{\ast }(\tau )g(\tau +t)\mathrm{d}\tau e^{-j2\pi ut}\mathrm{d}t\\ & =\int f^{\ast }(\tau )\int g(\tau +t)e^{-j2\pi ut}\mathrm{d}t\mathrm{d}\tau \\ & =G(u)\int f^{\ast }(\tau )e^{j2\pi u\tau }\mathrm{d}\tau \\ & =G(u)[\int f(\tau )e^{-j2\pi u\tau }\mathrm{d}\tau {]}^{\ast }\\ & =F^{\ast }(u)G(u)\end{array}$$

4. 乘积的证明和上面是一样的, 无非是作用于$\stackrel{ˇ}{}$上.

5. pass

6. 平移:
   $$(f(t+t_0){)}^{\hat{}}=\int f(t+t_0)e^{-j2\pi ut}\mathrm{d}t=e^{j2\pi u{t}_0}\int f(t+t_0)e^{-j2\pi u(t+t_0)}\mathrm{d}t=e^{j2\pi t_{0}u}F(u).$$

7. 逆:
   $$(f(-t){)}^{\hat{}}=\int f(-t)e^{-j2\pi ut}\mathrm{d}t=\int f(t)e^{j2\pi ut}\mathrm{d}t=F(-u).$$

8. $\hat{F}(t)=\int F(u)e^{j2\pi tu}\mathrm{d}u=(F(-u){)}^{\stackrel{ˇ}{}}(t)=f(-t)$.

9. scaling:
   $$(f(at){)}^{\hat{}}(u)={\int }_{-\infty }^{+\infty }f(at)e^{-j2\pi ut}\mathrm{d}t=\frac{1}{|a|}{\int }_{-\infty }^{+\infty }f(z)e^{-j2\pi \frac{u}{a}z}dz=\frac{1}{|a|}F(\frac{u}{a}).$$

10. 导函数:
    $$\begin{array}{ll}{f}^{\prime }(t)& =\frac{\mathrm{d}}{\mathrm{d}t}{\int }_{-\infty }^{+\infty }F(u)e^{j2\pi tu}\mathrm{d}u\\ & ={\int }_{-\infty }^{+\infty }F(u)\frac{\mathrm{d}}{\mathrm{d}t}{e}^{j2\pi tu}\mathrm{d}u\\ & ={\int }_{-\infty }^{+\infty }F(u)(j2\pi u)e^{j2\pi tu}\mathrm{d}u\\ & =(j2\pi uF(u){)}^{\stackrel{ˇ}{}}.\end{array}$$

11. 同上

12. 这个性质说明了, 对原图像加以旋转, 等价地在频域上加以同样的旋转.
    $$\begin{array}{ll}G(\mathbf{u})& =\int f(R\mathbf{t})e^{-j2\pi {\mathbf{u}}^T\mathbf{t}}\mathrm{d}\mathbf{t}\\ & =\int f(\mathbf{x})e^{-j2\pi (R\mathbf{u}{)}^T\mathbf{x}}|R^{-1}|\mathrm{d}\mathbf{x}\\ & =\int f(\mathbf{x})e^{-j2\pi (R\mathbf{u}{)}^T\mathbf{x}}\mathrm{d}\mathbf{x}\\ & =F(R\mathbf{u})\end{array}$$

对称性

主要指的:

1. 奇函数
   $$f(x)=-f(-x).$$

2. 偶函数
   $$f(x)=f(-x).$$

3. 共轭对称
   $$f^{\ast }(x)=f(-x).$$

考查$f(t)=r(t)+ji(t)$与$F(u)=R(u)+jI(u)$之间的关系:

	$\hat{}$	$\iff$	$\stackrel{ˇ}{}$
1	$f(t)=r(t)$	$\iff$	$F^{\ast }(u)=F(-u)$
2	$f(t)=r(t)$	$\iff$	$R(u)=R(-u),I(u)=-I(-u)$
3	$f(t)=ji(t)$	$\iff$	$F^{\ast }(-u)=-F(u)$
4	$f(t)=ji(t)$	$\iff$	$R(u)=-R(-u),I(u)=I(-u)$

	$\hat{}$	$\stackrel{\mathcal{F}}{\iff }$	$\stackrel{ˇ}{}$
5	$f(-t)\in \mathbb{R}$	$\stackrel{\mathcal{F}}{\iff }$	$F^{\ast }(u)\in \mathbb{C}$
6	$f(-t)\in \mathbb{C}$	$\stackrel{\mathcal{F}}{\iff }$	$F(-u)\in \mathbb{C}$
7	$f^{\ast }(t)\in \mathbb{C}$	$\stackrel{\mathcal{F}}{\iff }$	$F^{\ast }(-u)\in \mathbb{C}$
8	$f(t)\in \mathbb{R}$, 且$f(t)=f(-t)$	$\stackrel{\mathcal{F}}{\iff }$	$F(u)\in \mathbb{R}$, 且满足$F(u)=F(-u)$
9	$f(t)\in \mathbb{R}$, 且$f(t)=-f(-t)$	$\stackrel{\mathcal{F}}{\iff }$	$F(u)=jI(u)\in \mathbb{C}$, 且$F(u)=-F(-u)$
10	$f(t)=ji(t)\in \mathbb{C}$, 且$f(t)=f(-t)$	$\stackrel{\mathcal{F}}{\iff }$	$F(u)=jI(u)\in \mathbb{C}$, 且$F(u)=F(-u)$
11	$f(t)=ji(t)\in \mathbb{C}$, 且$f(t)=-f(-t)$	$\stackrel{\mathcal{F}}{\iff }$	$F(u)\in \mathbb{R}$, 且满足$F(u)=-F(-u)$
12	$f(t)\in \mathbb{C}$, 且$f(t)=f(-t)$	$\stackrel{\mathcal{F}}{\iff }$	$F(u)\in \mathbb{C}$, 且满足$F(u)=F(-u)$
13	$f(t)\in \mathbb{C}$, 且$f(t)=-f(-t)$	$\stackrel{\mathcal{F}}{\iff }$	$F(u)\in \mathbb{C}$, 且满足$F(u)=-F(-u)$

1. real ->($\Leftarrow$采用反证法即可)
   $$\begin{array}{ll}{F}^{\ast }(u)& =[\int f(t)e^{-j2\pi ut}\mathrm{d}t{]}^{\ast }\\ & =\int f^{\ast }(t)e^{j2\pi ut}\mathrm{d}t\\ & =\int f(t)e^{-j2\pi (-u)t}\mathrm{d}t\to f\ast (t)=r^{\ast }(t)=r(t)\\ & =F(-u).\end{array}$$

2. 由1易证.

3. imaginary ->
   $$\begin{array}{ll}{F}^{\ast }(-u)& =[\int f(t)e^{j2\pi ut}\mathrm{d}t{]}^{\ast }\\ & =\int f^{\ast }(t)e^{-j2\pi ut}\mathrm{d}t\\ & =\int -f(t)e^{-j2\pi ut}\mathrm{d}t\to f^{\ast }(t)=-ji(t)=-f(t)\\ & =-F(u).\end{array}$$

4. 由3易证.

5. 由性质6以及上述的对称性3可知$(f(-t){)}^{\hat{}}=F(-u)=F^{\ast }(u)$.

6. 由性质6可知$(f(-t){)}^{\hat{}}=F(-u)$.

7. $$\int f^{\ast }(t)e^{-j2\pi ut}\mathrm{d}t=[\int f(t)e^{j2\pi ut}\mathrm{d}t{]}^{\ast }=F^{\ast }(-u).$$

8. $$F(u)=(f(t){)}^{\hat{}}=(f(-t){)}^{\hat{}}=F(-u)=F^{\ast }(u).$$

9. $$F(u)=(f(t){)}^{\hat{}}=-(f(-t){)}^{\hat{}}=-F(-u)=-F^{\ast }(u).$$

10. $$F(u)=(f(t){)}^{\hat{}}=(f(-t){)}^{\hat{}}=F(-u)=-F^{\ast }(u).$$

11. $$F(u)=(f(t){)}^{\hat{}}=-(f(-t){)}^{\hat{}}=-F(-u)=F^{\ast }(u).$$

12. $$F(u)=(f(t){)}^{\hat{}}=(f(-t){)}^{\hat{}}=F(-u).$$

13. $$F(u)=(f(t){)}^{\hat{}}=-(f(-t){)}^{\hat{}}=-F(-u).$$

卷积

这里特别说明离散卷积的一个重要性质:
注意到
$$\begin{gathered} h\star f(n)=\sum _{k=0}^{N-1}h(n-k)f(k)=a_n^{T}f, \\ a(n)=[h(0),h(n-1),\cdots ,h(n-N+1)], \end{gathered}$$
则
$$A=[a_0,a_1,\cdots ,a_{N-1}]=\left[\begin{array}{cccc}h(0)& h(1)& \cdots & h(N-1)\\ h(-1)& h(0)& \cdots & h(N-2)\\ ⋮& ⋮& \ddots & ⋮\\ h(1-N)& h(2-N)& \cdots & h(0),\end{array}\right][h\star f]=A^{T}f.$$
定义
$$h^{\prime }(n)=h(-n),$$
则有
$$[h^{\prime }\star f]=Af.$$

FFT

考虑离散傅里叶变换:
$$F_m=\sum _{n=0}^{N-1}{f}_n{e}^{-j2\pi nm/N},m=0,1,\cdots ,N-1.$$

每次计算包含$N$个乘法, 故傅里叶变换的计算量是$N^2$级别的.

下面假设$N=P\ast Q,P,Q\in {\mathbb{N}}^{+}$, 则
$$\begin{gathered} m=uQ+v,u=0,1,\cdots ,P-1,v=0,1,\cdots ,Q-1, \\ n=sP+t,s=0,1,\cdots ,Q-1,t=0,1,\cdots ,P-1. \end{gathered}$$

注意到:
$$nm=(sP+t)(uQ+v)=suN+svP+tm.$$
此时, $m,n$与唯一的$(u,v),(s,t)$匹配, 则傅里叶变换可以表示为:
KaTeX parse error: No such environment: align at position 8: \begin{̲a̲l̲i̲g̲n̲}̲ F_m &= \sum_{n…

令
$$F_v^t:=\sum _{s=0}^{Q-1}{f}_{sP+t}{e}^{-j2\pi sv/Q},v=0,1,\cdots ,Q-1,t=0,1,\cdots ,P-1.$$
显然总共需要$QN$次乘法, 而
$$\begin{gathered} F_m=\sum _{t=0}^{P-1}{e}^{-j2\pi tm/N}{F}_v^t,m=0,1,\cdots ,N-1, \\ {F}_{uQ+v}=\sum _{t=0}^{P-1}{e}^{-j2\pi tv/N}{F}_v^t{e}^{-j2\pi tu/P}. \end{gathered}$$
总共有$PN$次乘法, 故总共有$(P+Q)N$次乘法运算.
由于每个$F^t$都是一个独立的傅里叶变换过程, 故倘若$Q$能够进一步分解, 计算量可以进一步降低.
甚至, 若假设$N=P^J$, 则可以证明最后的计算量可以分解为
$$PJN=PN{\log }_{P}N,$$
特别的, 常常$N=2^J$, 则计算量是$N{\log }_{2}N$量级的.

注: 卷积运算$[f\star g{]}_n,n=0,\cdots ,N-1$, 计算量也是$N^2$级别的, 但是通过FFT, 应当是$(PN{\log }_P+1)N$​乘法次数, 也是可以降低计算量的.

逆傅里叶变换实际上就是
$${\widehat{F^{\ast }}}^{\ast }=[\sum _{m=0}^{N-1}{F}_m^{\ast }{e}^{-j2\pi mn/N}{]}^{\ast },$$
虽然下面的代码我并没有采取这种策略.

from typing import Iterable, Union
import numpy as np



def factorization(N: int):
    for P in range(2, N + 1):
        if N % P is 0:
            return P

def dft(arr: Iterable, m: int):
    N = len(arr)
    basis = np.arange(N) * m * 2 * np.pi * 1j * -1
    basis = np.exp(basis / N)
    return arr @ basis

def idft(arr: Iterable, m: int):
    N = len(arr)
    basis = np.arange(N) * m * 2 * np.pi * 1j
    basis = np.exp(basis / N) / N
    return arr @ basis

def basis_dft(N: int):
    basis = np.outer(np.arange(N), np.arange(N))
    basis = basis * 2 * np.pi * 1j * -1
    return np.exp(basis / N)

def basis_idft(N: int):
    basis = np.outer(np.arange(N), np.arange(N))
    basis = basis * 2 * np.pi * 1j
    return np.exp(basis / N) / N

def fft(arr: Iterable) -> np.ndarray:
    N = len(arr)
    P = factorization(N)
    if P is N:
        return arr @ basis_dft(N)
    Q = N // P
    Fvt = np.array([fft(arr[t::P]) for t in range(P)]).T # Q x P
    dummy = np.outer(np.arange(Q), np.arange(P))
    dummy = np.exp((dummy * 2 * np.pi * 1j * -1) / N)
    Fvt_ext = Fvt * dummy
    return (Fvt_ext @ basis_dft(P)).T.flatten()

def ifft(arr: Iterable) -> np.ndarray:
    N = len(arr)
    P = factorization(N)
    if P is N:
        return arr @ basis_idft(N)
    Q = N // P
    Fvt = np.array([ifft(arr[t::P]) for t in range(P)]).T # Q x P
    dummy = np.outer(np.arange(Q), np.arange(P))
    dummy = np.exp((dummy * 2 * np.pi * 1j) / N)
    Fvt_ext = Fvt * dummy
    return (Fvt_ext @ basis_idft(P)).T.flatten()