Review DTFT DTFT Properties Examples Summary Lecture 9: Discrete-Time Fourier Transform Mark Hasegawa-Johnson ECE 401: Signal and Image Analysis, Fall 2020
Review DTFT DTFT Properties Examples Summary Review: Frequency Response 1 Discrete Time Fourier Transform 2 Properties of the DTFT 3 Examples 4 Summary 5
Review DTFT DTFT Properties Examples Summary Outline Review: Frequency Response 1 Discrete Time Fourier Transform 2 Properties of the DTFT 3 Examples 4 Summary 5
Review DTFT DTFT Properties Examples Summary What is Signal Processing, Really? When we process a signal, usually, we’re trying to enhance the meaningful part, and reduce the noise. Spectrum helps us to understand which part is meaningful, and which part is noise. Convolution (a.k.a. filtering) is the tool we use to perform the enhancement. Frequency Response of a filter tells us exactly which frequencies it will enhance, and which it will reduce.
Review DTFT DTFT Properties Examples Summary Review: Convolution A convolution is exactly the same thing as a weighted local average . We give it a special name, because we will use it very often. It’s defined as: � � y [ n ] = g [ m ] f [ n − m ] = g [ n − m ] f [ m ] m m We use the symbol ∗ to mean “convolution:” � � y [ n ] = g [ n ] ∗ f [ n ] = g [ m ] f [ n − m ] = g [ n − m ] f [ m ] m m
Review DTFT DTFT Properties Examples Summary Review: DFT & Fourier Series Any periodic signal with a period of N samples, x [ n + N ] = x [ n ], can be written as a weighted sum of pure tones, N − 1 x [ n ] = 1 � X [ k ] e j 2 π kn / N , N k =0 which is a special case of the spectrum for periodic signals: ω 0 = 2 π radians F 0 = 1 cycles T 0 = N seconds N = samples sample , second , cycle , cycle , N T 0 F s and N − 1 � x [ n ] e − j 2 π kn / N . X [ k ] = n =0
Review DTFT DTFT Properties Examples Summary Tones in → Tones out Suppose I have a periodic input signal, N − 1 x [ n ] = 1 � X [ k ] e j 2 π kn / N , N k =0 and I filter it, y [ n ] = h [ n ] ∗ x [ n ] , Then the output is a sum of pure tones, at the same frequencies as the input, but with different magnitudes and phases: N − 1 y [ n ] = 1 � Y [ k ] e j 2 π kn / N . N k =0
Review DTFT DTFT Properties Examples Summary Frequency Response Suppose we compute y [ n ] = x [ n ] ∗ h [ n ], where N − 1 x [ n ] = 1 � X [ k ] e j 2 π kn / N , and N k =0 N − 1 y [ n ] = 1 � Y [ k ] e j 2 π kn / N . N k =0 The relationship between Y [ k ] and X [ k ] is given by the frequency response: Y [ k ] = H ( k ω 0 ) X [ k ] where ∞ � h [ n ] e − j ω n H ( ω ) = n = −∞
Review DTFT DTFT Properties Examples Summary Outline Review: Frequency Response 1 Discrete Time Fourier Transform 2 Properties of the DTFT 3 Examples 4 Summary 5
Review DTFT DTFT Properties Examples Summary Aperiodic An “aperiodic signal” is a signal that is not periodic. Periodic acoustic signals usually have a perceptible pitch frequency; aperiodic signals sound like wind noise, or clicks. Music: strings, woodwinds, and brass are periodic, drums and rain sticks are aperiodic. Speech: vowels and nasals are periodic, plosives and fricatives are aperiodic. Images: stripes are periodic, clouds are aperiodic. Bioelectricity: heartbeat is periodic, muscle contractions are aperiodic.
Review DTFT DTFT Properties Examples Summary Periodic The spectrum of a periodic signal is given by its Fourier series, or equivalently in discrete time, by its discrete Fourier transform: N − 1 x [ n ] = 1 X [ k ] e j 2 π kn � N N k =0 N − 1 x [ n ] e − j 2 π kn � X [ k ] = N n =0
Review DTFT DTFT Properties Examples Summary Aperiodic The spectrum of an aperiodic signal we will now define to be exactly the same as that of a periodic signal except that, since it never repeats itself, its period has to be N = ∞ : N − 1 1 X [ k ] e j 2 π kn � x [ n ] ≈ lim N N N →∞ k =0 N − 1 x [ n ] e − j 2 π kn � X [ k ] ≈ lim N N →∞ n =0
Review DTFT DTFT Properties Examples Summary An Aperiodic Signal is like a Periodic Signal with Period= ∞
Review DTFT DTFT Properties Examples Summary Aperiodic The spectrum of an aperiodic signal we will now define to be exactly the same as that of a periodic signal except that, since it never repeats itself, its period has to be N = ∞ : N − 1 1 X [ k ] e j 2 π kn � x [ n ] ≈ lim N N N →∞ k =0 N − 1 x [ n ] e − j 2 π kn � X [ k ] ≈ lim N N →∞ n =0 But what does that mean? For example, what is 2 π k N ? Let’s try this definition: allow k → ∞ , and force ω to remain constant, where ω = 2 π k N
Review DTFT DTFT Properties Examples Summary Aperiodic Let’s start with this one: N − 1 1 X [ k ] e j 2 π kn � x [ n ] ≈ lim N N N →∞ k =0 Imagine this as adding up a bunch of tall, thin rectangles, each with a height of X [ k ], and a width of d ω = 2 π N . In the limit, as N → ∞ , that becomes an integral: N − 1 1 2 π N X [ k ] e j 2 π kn � x [ n ] ≈ lim N 2 π N →∞ k =0 � 2 π = 1 X ( ω ) e j ω n d ω, 2 π ω =0 where we’ve used X ( ω ) = X [ k ] just because, as k → ∞ , it makes more sense to talk about X ( ω ).
Review DTFT DTFT Properties Examples Summary Approximating the Integral as a Sum
Review DTFT DTFT Properties Examples Summary Periodic Now, let’s go back to periodic signals. Notice that e j 2 π = 1, and for that reason, e j 2 π k ( n + N ) = e j 2 π k ( n − N ) = e j 2 π kn N . So in the DFT, we N N get exactly the same result by summing over any complete period of the signal: N − 1 x [ n ] e − j 2 π kn � X [ k ] = N n =0 N x [ n ] e − j 2 π kn � = N n =1 N − 4 x [ n ] e − j 2 π kn � = N n = − 3 N − 1 2 x [ n ] e − j 2 π kn � = N n = − ( N − 1) 2 2 π kn �
Review DTFT DTFT Properties Examples Summary Aperiodic Let’s use this version, because it has a well-defined limit as N → ∞ : N − 1 2 x [ n ] e − j 2 π kn � X [ k ] = N n = − ( N − 1) 2 The limit is: N − 1 2 � x [ n ] e − j ω n X ( ω ) = lim N →∞ n = − ( N − 1) 2 ∞ � x [ n ] e − j ω n = n = −∞
Review DTFT DTFT Properties Examples Summary Discrete Time Fourier Transform (DTFT) So in the limit as N → ∞ , � π x [ n ] = 1 X ( ω ) e j ω n d ω 2 π − π ∞ � x [ n ] e − j ω n X ( ω ) = n = −∞ X ( ω ) is called the discrete time Fourier transform (DTFT) of the aperiodic signal x [ n ].
Review DTFT DTFT Properties Examples Summary Outline Review: Frequency Response 1 Discrete Time Fourier Transform 2 Properties of the DTFT 3 Examples 4 Summary 5
Review DTFT DTFT Properties Examples Summary Properties of the DTFT In order to better understand the DTFT, let’s discuss these properties: 0 Periodicity 1 Linearity 2 Time Shift 3 Frequency Shift 4 Filtering is Convolution Property #4 is actually the reason why we invented the DTFT in the first place. Before we discuss it, though, let’s talk about the others.
Review DTFT DTFT Properties Examples Summary 0. Periodicity The DTFT is periodic with a period of 2 π . That’s just because e j 2 π = 1: � x [ n ] e − j ω n X ( ω ) = n x [ n ] e − j ( ω +2 π ) n = x [ n ] e − j ω n = X ( ω ) � � X ( ω + 2 π ) = n n x [ n ] e − j ( ω − 2 π ) n = x [ n ] e − j ω n = X ( ω ) � � X ( ω − 2 π ) = n n In fact, we’ve already used this fact. I defined the inverse DTFT in two different ways: � π � 2 π x [ n ] = 1 X ( ω ) e j ω n d ω = 1 X ( ω ) e j ω n d ω 2 π 2 π − π 0 Those two integrals are equal because X ( ω + 2 π ) = X ( ω ).
Review DTFT DTFT Properties Examples Summary 1. Linearity The DTFT is linear: z [ n ] = ax [ n ] + by [ n ] Z ( ω ) = aX ( ω ) + bY ( ω ) ↔ Proof: � z [ n ] e − j ω n Z ( ω ) = n x [ n ] e − j ω n + b � � y [ n ] e − j ω n = a n n = aX ( ω ) + bY ( ω )
Review DTFT DTFT Properties Examples Summary 2. Time Shift Property Shifting in time is the same as multiplying by a complex exponential in frequency: Z ( ω ) = e − j ω n 0 X ( ω ) z [ n ] = x [ n − n 0 ] ↔ Proof: ∞ � x [ n − n 0 ] e − j ω n Z ( ω ) = n = −∞ ∞ � x [ m ] e − j ω ( m + n 0 ) = (where m = n − n 0 ) m = −∞ = e − j ω n 0 X ( ω )
Review DTFT DTFT Properties Examples Summary 3. Frequency Shift Property Shifting in frequency is the same as multiplying by a complex exponential in time: z [ n ] = x [ n ] e j ω 0 n Z ( ω ) = X ( ω − ω 0 ) ↔ Proof: ∞ � x [ n ] e j ω 0 n e − j ω n Z ( ω ) = n = −∞ ∞ � x [ n ] e − j ( ω − ω 0 ) n = n = −∞ = X ( ω − ω 0 )
Recommend
More recommend
