dsp first, 2/e
TRANSCRIPT
![Page 1: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/1.jpg)
DSP First, 2/e
Lecture 9
Sampling & Aliasing
MODIFIED TLH
![Page 2: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/2.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 2
Let’s Go to the MoviesSAMPLING AND ALIASING REFERENCES CH 4 2/12/2019
Interesting illusion: the helicopter's blades are somehow in sync with
the camera's shutter making it seem as though they are not moving.
0:47 https://www.youtube.com/watch?v=qgvuQGY946g
Helicopter blades and other fast spinning objects often produce
strange effects on camera. Lauren explains why. 3:51https://www.youtube.com/watch?v=AYQAKwCxScc
Audio Sampling Rate Demo 0:35
https://www.youtube.com/watch?v=hRhVb6iRArg&feature=youtu.be
![Page 3: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/3.jpg)
Video Aliasing
May 2016 © 2003-2016, JH McClellan & RW Schafer 3
Why car wheels rotate
backwards in movies 4:25https://www.youtube.com/watch?v=SFbINinFsxk&feature=yout
u.be
![Page 4: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/4.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 4
READING ASSIGNMENTS
This Lecture:
Chap 4, Sections 4-1 and 4-2
Other Reading:
TLH WEBSITE
Next Lecture: Chap. 4, Sects. 4-3 and 4-4
![Page 5: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/5.jpg)
Sampling
Figure 4-1: Block diagram representation of the ideal continuous-to-discrete (C-to-D)
converter. The parametersT specifies uniform sampling of the input signal every
sT seconds.
Sometimes ADC, A2D, Analog-to-Digital
n = 1,2,3,4… nTs in seconds
![Page 6: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/6.jpg)
Copyright © 2016, 1998 Pearson Education, Inc. All Rights Reserved
Sampling Sinusoidal Signals (1 of 2)
Figure 4-2: Plotting format for discrete-time signals, called a stem plot. In MATLAB, the
function stem produces this plot. Some students also refer to the stem plot as a “lollypop”
plot.
![Page 7: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/7.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 7
>> help stem
stem Discrete sequence or "stem" plot.
stem(Y) plots the data sequence Y as stems from the x axis
terminated with circles for the data value. If Y is a matrix
then each column is plotted as a separate series.
stem(X,Y) plots the data sequence Y at the values specified
in X.
![Page 8: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/8.jpg)
Copyright © 2016, 1998 Pearson Education, Inc. All Rights Reserved
Sampling Sinusoidal Signals (2 of 2)
Figure 4-3: A continuous-time 100 Hz sinusoid (a) and two discrete-time sinusoids
formed by sampling at samples/s 2000sf (b) and at (c). samples/s 500sf
![Page 9: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/9.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 9
LECTURE OBJECTIVES
SAMPLING can cause ALIASING
Sampling Theorem
Sampling Rate > 2(Highest Frequency)
Spectrum for digital signals, x[n]
Normalized Frequency
22
ˆ s
sf
fT
ALIASING
![Page 10: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/10.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 10
SYSTEMS Process Signals
PROCESSING GOALS:
Change x(t) into y(t)
For example, more BASS, pitch shifting
Improve x(t), e.g., image deblurring
Extract Information from x(t)
SYSTEMx(t) y(t)
![Page 11: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/11.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 11
System IMPLEMENTATION
DIGITAL/MICROPROCESSOR Convert x(t) to numbers stored in memory
ELECTRONICSx(t) y(t)
COMPUTER D-to-AA-to-Dx(t) y(t)y[n]x[n]
ANALOG/ELECTRONIC: Circuits: resistors, capacitors, op-amps
![Page 12: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/12.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 12
SAMPLING x(t)
SAMPLING PROCESS Convert x(t) to numbers x[n]
“n” is an integer index; x[n] is a sequence of values
Think of “n” as the storage address in memory
UNIFORM SAMPLING at t = nTs
IDEAL: x[n] = x(nTs)
C-to-Dx(t) x[n]
![Page 13: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/13.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 13
SAMPLING RATE, fs
SAMPLING RATE (fs)
fs =1/Ts
NUMBER of SAMPLES PER SECOND
Ts = 125 microsec fs = 8000 samples/sec• UNITS of fs ARE HERTZ: 8000 Hz
UNIFORM SAMPLING at t = nTs = n/fs IDEAL: x[n] = x(nTs)=x(n/fs)
C-to-Dx(t) x[n]=x(nTs)
![Page 14: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/14.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 14
STORING DIGITAL SOUND
x[n] is a SAMPLED SISIGNAL
A list of numbers stored in memory
EXAMPLE: audio CD
CD rate is 44,100 samples per second
16-bit samples
Stereo uses 2 channels
Number of bytes for 1 minute is
2 X (16/8) X 60 X 44100 = 10.584 Mbytes
![Page 15: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/15.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 15Which one provides the most accurate representation of x(t)?
fs 2 kHz
fs 500Hz
Hz100f
![Page 16: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/16.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 16
SAMPLING THEOREM
HOW OFTEN DO WE NEED TO SAMPLE?
DEPENDS on FREQUENCY of SINUSOID
ANSWERED by SHANNON/NYQUIST Theorem
ALSO DEPENDS on “RECONSTRUCTION”
![Page 17: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/17.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 17
Reconstruction? Which One?
)4.0cos(][ nnx
When n is an integer
cos(0.4n) cos(2.4n)
Given the samples, draw a sinusoid through the values
Occam’s razor -> pick lowest frequency sinusoid
![Page 18: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/18.jpg)
Be careful…https://www.youtube.com/watch?v=qgvuQGY946g
May 2016 © 2003-2016, JH McClellan & RW Schafer 18
See the References on the Website
![Page 19: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/19.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 19
sfsT
nAnx
ˆ
)ˆcos(][
)cos()(][
)cos()(
ss nTAnTxnx
tAtx
DISCRETE-TIME SINUSOID
Change x(t) into x[n] DERIVATION
))cos((][ nTAnx s
DEFINE DIGITAL FREQUENCY
![Page 20: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/20.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 20
DIGITAL FREQUENCY
VARIES from 0 to 2, as f varies from
0 to the sampling frequency
UNITS are radians, not rad/sec
DIGITAL FREQUENCY is NORMALIZED
ˆ Ts 2f
fs
ˆ
![Page 21: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/21.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 21
SPECTRUM (DIGITAL)
sf
f 2ˆ
kHz1sf
12 X1
2 X*
2(0.1)–0.2
))1000/)(100(2cos(][ nAnx
![Page 22: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/22.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 22
SPECTRUM (DIGITAL) ???
ˆ
12 X1
2 X*
2(1)–2
?
x[n] is zero frequency???
))100/)(100(2cos(][ nAnx
sf
f 2ˆ
fs 100 Hz
![Page 23: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/23.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 23
rad, 52ˆ
samples/s. 80sf100 Hz sinusoid
![Page 24: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/24.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 24
The REST of the STORY
Spectrum of x[n] has more than one line for
each complex exponential
Called ALIASING
MANY SPECTRAL LINES
SPECTRUM is PERIODIC with period = 2 Because
( ) ( )( ) nAnA 2ˆcosˆcos
![Page 25: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/25.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 25
ALIASING DERIVATION
Other Frequencies give the same
Hz1000at sampled)400cos()(1 sfttx
)4.0cos()400cos(][10001 nnx n
Hz1000at sampled)2400cos()(2 sfttx
)4.2cos()2400cos(][10002 nnx n
)4.0cos()24.0cos()4.2cos(][2 nnnnnx
][][ 12 nxnx )1000(24002400
ˆ
![Page 26: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/26.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 26
ALIASING DERIVATION
Other Frequencies give the same
s
sf
fT
2ˆ 2
ˆ
s
s
ss
s
f
f
f
f
f
ff
22)(2ˆ :then
and we want : x[n] Acos( ˆ n )
))(2cos()( If tffAtx s
tn
fs
![Page 27: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/27.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 27
ALIASING CONCLUSIONS
Adding an INTEGER multiple of fs or –fs to
the frequency of a continuous sinusoid xc(t)
gives exactly the same values for the
sampled signal x[n] = xc(n/fs )
GIVEN x[n], we CAN’T KNOW whether it came
from a sinusoid at fo or (fo + fs ) or (fo + 2fs ) …
This is called ALIASING
![Page 28: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/28.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 28
SPECTRUM for x[n]
PLOT versus NORMALIZED FREQUENCY
INCLUDE ALL SPECTRUM LINES
ALIASES
ADD MULTIPLES of 2
SUBTRACT MULTIPLES of 2
FOLDED ALIASES
(to be discussed later)
ALIASES of NEGATIVE FREQS
![Page 29: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/29.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 29
SPECTRUM (MORE LINES)
12 X1
2 X*
2(0.1)–0.2
12 X
*
1.8
12 X
–1.8
))1000/)(100(2cos(][ nAnx
kHz1sf
sf
f 2ˆ
![Page 30: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/30.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 30
SPECTRUM (ALIASING CASE)
12 X
*
–0.5
12 X
–1.5
12 X
0.5 2.5–2.5
12 X1
2 X* 1
2 X*
1.5
))80/)(100(2cos(][ nAnx
kHz80sf
sf
f 2ˆ
![Page 31: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/31.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 31
SAMPLING GUI (con2dis)
![Page 32: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/32.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 32
SAMPLING GUI (con2dis)
![Page 33: DSP First, 2/e](https://reader031.vdocuments.site/reader031/viewer/2022012517/619116ecb27b5f219e2fb1c3/html5/thumbnails/33.jpg)
May 2016 © 2003-2016, JH McClellan & RW Schafer 33
SPECTRUM (FOLDING CASE)
2f
fs
fs 125Hz
12 X
*
0.4
12 X
–0.4 1.6–1.6
12 X1
2 X*
))125/)(100(2cos(][ nAnx