Scribd
Upload
40 views

Matlab Tutorial I

The document discusses four projects on adaptive signal processing topics: 1) denoising a noisy digital speech signal, 2) denoising a noisy digital image, 3) determining the power spectral density of a random signal/process, and 4) analyzing the output of a linear system under a stationary random process input. Students are instructed to complete simulations and analysis for each project in MATLAB.

Uploaded by

SatwikMohanty
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
40 views

Matlab Tutorial I

The document discusses four projects on adaptive signal processing topics: 1) denoising a noisy digital speech signal, 2) denoising a noisy digital image, 3) determining the power spectral density of a random signal/process, and 4) analyzing the output of a linear system under a stationary random process input. Students are instructed to complete simulations and analysis for each project in MATLAB.

Uploaded by

SatwikMohanty
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

MATLAB Tutorial for Adaptive Signal Processing-I

SUBJECT: Adaptive Signal Processing SUBJECT CODE: EE-645


Venue: Embedded System and Real-Time Laboratory, EE Dept.
Semester: Autumn, 2017-18

Topics Covered:-
 Random Variable (RV), Random Vector (RVec), and Random/Stochastic Process (RP)
 Generation
 Ensemble Characteristics: Expectation, Variance, Auto-correlation, Power Spectral Density
 Denoising

Project 1: Noisy Digital Speech Signal


Instructions:

1. Get/Read a speech signal (download from the Internet or record your voice microphone) into
MATLAB and plot it.
2. Denote it as the speech signal v ( ).
3. Generate zero-mean Gaussian random noise η( ) using MATLAB function normrand. Assume
as last two digits or any combination of two digits from last four digits of your roll number in
case the last two digits are zeros.
4. Add η( ) to v ( ) (which may be induced in an speech recording system). Denote the resultant
noisy signal as s( ), that is, s( ) = v ( ) + η( ) .
5. Generate a set of = 3, 10, 20, 50, 100 noisy images, {s ( ), = 1,2, … , }.
6. Compute ensemble average speech signal s̅ (
) and variance for each and show it in figures.
7. Measure Mean Square Error (MSE) between s(t), noisy signals, {s ( ), = 1,2, … , }

What is your interpretation of MSE results? Provide some theoretical analysis. (Hint: determine the
mean and variance of s(t) and ̅ (t).

Project 2: Noisy Digital Image


Instructions:
1. Get a gray scale image (download from the Internet or take it using your own digital camera).
Denote it as the original image I ( ,
), where ( , ) is a pair of coordinates of a pixel.
2. Generate zero-mean Gaussian random noise η( , ) using MATLAB function normrand.
Assume as last two digits or any combination of two digits from last four digits of your roll
number in case the last two digits are zeros.
3. Add η( , ) to I ( , ) (which may be induced in an image acquisition system). Denote the
resultant noisy image as g( , ), that is, g( , ) = I ( , ) + η( , ) .
4. Generate a set of = 3, 10, 20, 50, 100 noisy images, {g ( , ), = 1,2, … , }.
5. Compute ensemble average image g( , ) and variance for each and show it in figures.
6. Measure Mean Square Error (MSE) between g(x, y), noisy images, {g ( , ), = 1,2, … , }
What is your interpretation of MSE results? Provide some theoretical analysis. (Hint: determine the
mean and variance of g(x, y) and g(x, y).

Project 3: Power Spectral Density of Random Signal / Process


1. Generate a random process ( ) = (0.1 2 ) + 1.5 (1,32) with = [0 1 2 … 31].
2. Generate a set of = 100 realizations, {x ( ), = 1,2, … , }.
3. Use ‘xcorr’ MATLAB function to determine correlation per realization and do ensemble averaging to get
auto-correlation sequence ( ) of the random process
4. Plot ( ) and check if its properties for WSS process hold
5. Obtain the Power Spectral Density (PSD) of the given process using DFT of ( ) with appropriate length
and check its properties. What is the area under the PSD curve?

Project 4: Linear System under Stationary Random Process


1. Generate a random process ( ) = 0.85 ( 1) + ( ) when the input is WN with variance 4 and
plot it.
2. A systems with a transfer functions ( ) = ( . )
is excited by the ( ).
3. Obtain the Power Spectral Density (PSD) of the output process of the given system.
4. Determine the PSD of overall system and compare with what you got in above step.

You might also like