[EE261学习笔记] 14.矩阵形式的离散傅里叶变换,及离散傅里叶变换的相关公式

本文主要罗列一下DFT的相关公式


首先,不难将DFT写成矩阵形式,回忆之前的推导,DFT表达式为:

F f ‾ = ∑ n = 0 N − 1 f [ n ] ‾ ω ‾ − n \underline{\mathscr{F}f} = \sum_{n=0}^{N-1} \underline{f[n]} \underline{\omega}^{-n} Ff=n=0N1f[n]ωn

写成矩阵形式就是:

[ F f [ 0 ] ‾ F f [ 1 ] ‾ ⋮ F f [ N − 1 ] ‾ ] = [ 1 1 ⋯ 1 1 ω ‾ − 1 ⋅ 1 ⋯ ω ‾ − 1 ⋅ ( N − 1 ) ⋮ ⋮ ⋱ ⋮ 1 ω ‾ − ( N − 1 ) ⋅ 1 ⋯ ω ‾ − ( N − 1 ) 2 ] [ f [ 0 ] ‾ f [ 1 ] ‾ ⋮ f [ N − 1 ] ‾ ] \large\begin{bmatrix}\underline{\mathscr{F}f[0]}\\\underline{\mathscr{F}f[1]}\\\vdots\\\underline{\mathscr{F}f[N-1]}\end{bmatrix} = \begin{bmatrix}1 & 1 & \cdots & 1 \\1 & \underline{\omega}^{-1 \cdot 1} & \cdots & \underline{\omega}^{- 1 \cdot (N-1)} \\\vdots & \vdots & \ddots & \vdots \\1 & \underline{\omega}^{-(N-1) \cdot 1} & \cdots & \underline{\omega}^{-(N-1)^2}\end{bmatrix}\begin{bmatrix}\underline{f[0] } \\\underline{f[1]} \\\vdots\\\underline{f[N-1]} \end{bmatrix} Ff[0]Ff[1]Ff[N1]=1111ω11ω(N1)11ω1(N1)ω(N1)2f[0]f[1]f[N1]

我们发现,DFT其实是一个对称矩阵,即

F ‾ T = F ‾ (1) \huge\underline\mathscr{F}^T = \underline\mathscr{F}\tag1 FT=F(1)

根据上一章中的运算,还可以知道:

F ‾ − 1 F ‾ = F ‾ F ‾ − 1 = N I (2) \huge\underline\mathscr{F}^{-1} \underline\mathscr{F}= \underline\mathscr{F} \underline\mathscr{F}^{-1} = NI\tag2 F1F=FF1=NI(2)

结合 ( 1 ) (1) (1) ( 2 ) (2) (2)两式,有:

F ‾ ∗ F ‾ = F ‾ F ‾ ∗ = N I \huge\underline\mathscr{F}^* \underline\mathscr{F}= \underline\mathscr{F} \underline\mathscr{F}^* = NI FF=FF=NI


DFT的对偶性:

  1. f ‾ − [ n ] = f ‾ [ − n ] \huge\underline{f}^- [n] = \underline{f}[-n] f[n]=f[n]

  2. F ‾ ( f ‾ − ) = ( F f ‾ ) − \huge\underline\mathscr{F}\left( \underline{f}^- \right) = \left(\underline{\mathscr{F} f} \right)^- F(f)=(Ff)

  3. F F f ‾ = N f − \huge\underline{\mathscr{F}\mathscr{F}f} = N f^- FFf=Nf

( 3 ) (3) (3) 也是与连续傅里叶变换情况不同的地方


类似连续情况,离散情况下的两个常用信号

a. 1 ‾ = ( 1 , 1 , … , 1 ) \underline{1} = (1, 1, \dots, 1) 1=(1,1,,1) 表示各处均为1的离散信号

b. δ k ‾ = ( 0 , … , 0 , 1 , 0 , … , 0 ) \underline{\delta_k} = (0, \dots, 0, 1, 0, \dots, 0) δk=(0,,0,1,0,,0) 表示 k k k 处为 1 1 1,其余均为 0 0 0 的离散信号

接下来我们来推导几个常用的DFT公式

  1. F δ k ‾ \underline{\mathscr{F}\delta_k} Fδk

F δ k ‾ = ∑ n = 0 N − 1 δ k [ n ] ‾ ω ‾ − n = 1 ⋅ ω ‾ − k = ω ‾ − k \begin{aligned} \underline{\mathscr{F}\delta_k} &= \sum_{n=0}^{N-1} \underline{\delta_k [n]} \underline{\omega}^{-n}\\ &= 1\cdot \underline{\omega}^{-k}\\ &= \underline{\omega}^{-k} \end{aligned} Fδk=n=0N1δk[n]ωn=1ωk=ωk

特别地, F δ ‾ = 1 ‾ \underline{\mathscr{F}\delta}=\underline{1} Fδ=1

  1. F ω k ‾ \underline{\mathscr{F}\omega^k} Fωk

F ω k ‾ = ∑ n = 0 N − 1 ω [ n ] ‾ k ω ‾ − n \underline{\mathscr{F}\omega^k} = \sum_{n=0}^{N-1} \underline{\omega [n]}^k \underline{\omega}^{-n} Fωk=n=0N1ω[n]kωn

我们看它的第 m m m项,

F ω [ m ] ‾ k = ∑ n = 0 N − 1 ω [ n ] ‾ k ω [ m ] ‾ − n = ∑ n = 0 N − 1 e 2 π i n k N e − 2 π i m n N = ∑ n = 0 N − 1 e 2 π i n ( k − m ) N = { 0 k ≠ m N k = m \begin{aligned} \underline{\mathscr{F}\omega[m]}^k &= \sum_{n=0}^{N-1} \underline{\omega [n]}^k \underline{\omega[m]}^{-n}\\ &= \sum_{n=0}^{N-1} e^{2\pi i\frac{nk}{N}} e^{-2\pi i\frac{mn}{N}}\\ &= \sum_{n=0}^{N-1} e^{2\pi i\frac{n(k-m)}{N}}\\ &= \begin{cases} 0 & k \neq m\\ N & k=m \end{cases} \end{aligned} Fω[m]k=n=0N1ω[n]kω[m]n=n=0N1e2πiNnke2πiNmn=n=0N1e2πiNn(km)={0Nk=mk=m

也就是说 F ω k ‾ \underline{\mathscr{F}\omega^k} Fωk 只有在 m = k m=k m=k 的时候值为 N N N,其他地方值均为 0 0 0,因此

F ω k ‾ = N δ k ‾ \underline{\mathscr{F}\omega^k} = N \underline{\delta^k} Fωk=Nδk

你可能感兴趣的:(学习笔记-傅里叶变换)