The difference and connection of DFT, DTFT, DFS/DTFS
Fourier transform can transform discrete sequences and continuous signals.
Fourier transform and Fourier series
CTFT: x ( t ) → X ( j Ω ) x(t)→X(j\Omega) x(t)→X ( j Ω )
transforms the continuous signal from the time domain to the frequency domain, the frequency domain is continuous, inΩ (rad / s) \Omega (rad/s)Ω ( r a d / s ) is the coordinate axis, draw [0,2π], the
computer cannot do this transformation, it can only be derived mathematically, or sample a continuous signal into a discrete sequence, and use FFT to increase the number of points N To approximate CTFT
DTFT(FT): x ( n ) → X ( e j ω ) x(n)→X(e^{j\omega}) x(n)→X ( andj ω )
transforms the discrete sequence from the time domain to the frequency domain. The frequency domain is continuous, withω (rad) \omega(rad)ω ( r a d ) is the coordinate axis. Drawing [0,2π]
DTFT uses less, because continuous frequency data cannot be processed by a computer
Divided into two groups of memory is more convenient, about the frequency coordinate relationship f, Ω, ω, kf,\Omega,\omega,kf,Ω ,ω ,k can refer tothe relationship between physical frequency, circular frequency, and digital frequency
DFT: x ( n ) → X ( k ) x(n)→X(k) x(n)→X ( k )
transforms the discrete sequence from the time domain to the frequency domain. The frequency domain is discrete, withkkk is the coordinate axis, draw [0,N]
when using DFT function (or fast Fourier transform FFT), it is necessary to determine the number of transformation points NNN , generally take2 n 2^n2n, k k The maximum value of k is N, which corresponds to 2π in DTFT
%matlab程序
N = 512;
xn = [1 2 3 0 0 0];
Xk = fft(xn,N)
DFS/DTFS: x ~ ( n ) → X ~ ( k ) \tilde{x}(n)→\tilde{X}(k) x~(n)→X~ (k)
Transform the infinitely long sequence from the time domain to the frequency domain. The frequency domain is discrete, withkkk is the coordinate axis
if x ~ (n) \tilde(x)(n)x~ (n)isx (n) x(n)The period extension of x ( n ) , thenX ~ (k) \tilde{X}(k)X~ (k)is alsoX (k) X(k)X ( k ) period extension, so DFT and DFS are actuallyTake the main valueversusCycle extensionRelationship
Supplement: DFT is equivalent to sampling DTFT