DEPARTMENT OF ELECTRONICS AND COMMUNICATION ENGINEERING

DIGITAL IMAGE PROCESSING

LAB MANUAL

Program/ Branch : B. E., / ECE

Year / Semester : III/ V
Academic Year : 2021 – 2022 (Odd Semester)
Regulation : R 2017

HOD/ ECE
 Category L T P Credit
17ECCC87 DIGITAL IMAGE PROCESSING LAB
CC 0 0 4 2
PREAMBLE
To understand and implement image processing techniques using open source software
PRERQUISITE
Signal Processing
COURSE OBJECTIVES
1 To understand image acquisition and storage using a open source software – SCILAB
2 To study and analyze different image transforms on images
To study, analyze and apply different techniques and algorithms for image
3
enhancement
To study, analyze and apply different techniques and algorithms for image
4
restoration
To study, analyze and apply different techniques and algorithms for image
5
compression
COURSE OUTCOMES
On the successful completion of the course, students will be able to
CO1.Understand the acquisition and storage of different types of images Understand
CO2. Understand Analyze and Apply different image transforms and
Apply
their properties
CO3. Apply different Image Smoothening & Sharpening algorithms in
Apply
time and frequency domain
CO4. Apply different algorithms for image restoration Apply
CO5. Apply different techniques for image segmentation Apply
CO6. Apply different image compression techniques Apply
MAPPING WITH PROGRAMME OUTCOMES AND PROGRAMME SPECIFIC OUTCOMES
PO PO PO PO PO PO PO PO PO PO PO PO
COS PSO1 PSO2 PSO3
1 2 3 4 5 6 7 8 9 10 11 12
CO1 S M - - M - - - - - - M - - -
CO2 S M M M M - - - - - - M - - -
CO3 S M M M M - - - - - - M - - -
CO4 S M M M M - - - - - - M - - -
CO5 S M M M M - - - - - - M - - -
CO6 S M M M M - - - - - - M - - -
S- Strong; M-Medium; L-Low
List of Experiments

1) To acquire an image, store in different formats and display the properties of the
images
2) To find the discrete Fourier transform of a gray scale image and perform inverse
transform to get back the image
3) Analyze the rotation and convolution properties of the Fourier transform using any
gray scale image
4) Find the discrete cosine transform of a given image. Compare discrete Fourier
transform and discrete cosine transforms
5) Apply histogram equalization for enhancing the given images
6) Perform image enhancement, smoothing and sharpening, in spatial domain using
different spatial filters and compare the performances
7) Perform image enhancement, smoothing and sharpening, in frequency domain
using different filters and compare the performances
8) Perform noise removal using different spatial filters and compare their
performances
9) For the given image perform edge detection using different operators and compare
the results
10)For a given image, compress and decompress using wavelets. Study and compare
the efficiency of the scheme with any two schemes

COURSE DESIGNERS

S. No. Name of the Faculty Mail ID

1 Mr. P. Subramanian [email protected]

2 Mrs.S.Valarmathy [email protected]

3 Mr.R.Ramani [email protected]
 1. IMAGE TYPES
Aim:
To display images of different types along with the information about the
images

Theory:
The different image types used in this program are
bmp – bitmap image file
jpeg – joint photographic expert group
png – portable network graphics
tiff – tagged Image File format

The functions used in this program are

subplot divides the current figure into rectangular panes that are numbered
rowwise. Each pane contains an axes object which you can manipulate using
Axes Properties. Subsequent plots are output to the current pane. h =
subplot(m,n,p) or subplot(mnp) breaks the figure window into an m-by-n
matrix of small axes, selects the pth axes object for the current plot, and
returns the axes handle. The axes are counted along the top row of the figure
window, then the second row, etc.

imshow – The function imshow(filename) displays the image stored in the

graphics file filename. The file must contain an image that can be read by
imread or dicomread. imshow calls imread or dicomread to read the image
from the file, but does not store the image data in the MATLAB workspace. If
the file contains multiple images, imshow displays the first image in the file.
The file must be in the current directory or on the MATLAB path.
impixelinfo – The function impixelinfo creates a Pixel Information tool in the
current figure. The Pixel Information tool displays information about the pixel
in an image that the pointer is positioned over. The tool can display pixel
information for all the images in a figure.
imageinfo – The function imageinfo creates an Image Information tool
associated with the image in the current figure. The tool displays information
about the basic attributes of the target image in a separate figure.
title – The function title('string') outputs the string at the top and in the center
of the current axes.
 MATLAB Program

clc;
clear all;
close all;

subplot(2,2,1), imshow('cameraman.tif'),title('cameraman.tif');
subplot(2,2,2), imshow('peppers.png'),title('peppers.png');
subplot(2,2,3), imshow('baby.bmp'),title('baby.bmp');
subplot(2,2,4), imshow('oldman.jpg'),title('oldman.png');
impixelinfo;

imageinfo('cameraman.tif');
imageinfo('peppers.png');
imageinfo('baby.bmp');
imageinfo('oldman.jpg');
 OUTPUT

Image Information for peppers.png

C:\Program
Filename
Files\MATLAB\R2009b\toolbox\images\imdemos\peppers.png
FileModDate 16-Dec-2002 06:10:58
FileSize 287677
Format png
FormatVersion []
Width 512
Height 384
BitDepth 24
ColorType truecolor
FormatSignature [137 80 78 71 13 10 26 10]
Colormap []
Histogram []
InterlaceType none
Transparency none
SimpleTransparencyData []
BackgroundColor []
RenderingIntent []
Chromaticities []
Gamma []
XResolution []
YResolution []
ResolutionUnit []
XOffset []
YOffset []
OffsetUnit []
SignificantBits []
ImageModTime 16 Jul 2002 16:46:41 +0000
Title []
Author []
Description Zesty peppers
Copyright Copyright The MathWorks, Inc.
CreationTime []
Software []
Disclaimer []
Warning []
Source []
Comment []
OtherText []

Image information for camearman.tif

C:\Program
Filename Files\MATLAB\R2009b\toolbox\images\imdemos\cameraman.
tif
FileModDate 04-Dec-2000 13:57:54
FileSize 65240
Format tif
FormatVersion []
Width 256
Height 256
BitDepth 8
ColorType grayscale
FormatSignature [77 77 42 0]
ByteOrder little-endian
NewSubFileType 0
BitsPerSample 8
Compression PackBits
PhotometricInterpretatio
BlackIsZero
n
StripOffsets [8x1 double]
SamplesPerPixel 1
RowsPerStrip 32
StripByteCounts [8x1 double]
XResolution 72
YResolution 72
ResolutionUnit None
Colormap []
PlanarConfiguration Chunky
TileWidth []
TileLength []
TileOffsets []
TileByteCounts []
Orientation 1
FillOrder 1
GrayResponseUnit 0.0100
MaxSampleValue 255
MinSampleValue 0
Thresholding 1
Offset 64872
This image is distributed by The MathWorks, Inc. with
ImageDescription
permission from the Massachusetts Institute of Technology.

Image information for baby.bmp

Filename E:\ baby.bmp
FileModDate 03-Mar-2006 04:05:44
FileSize 687054
Format Bmp
FormatVersion Version 3 (Microsoft Windows 3.x)
Width 500
Height 458
BitDepth 24
ColorType truecolor
FormatSignature BM
NumColormapEntries 0
Colormap []
RedMask []
GreenMask []
BlueMask []
ImageDataOffset 54
BitmapHeaderSize 40
NumPlanes 1
CompressionType None
BitmapSize 687000
HorzResolution 0
VertResolution 0
NumColorsUsed 0
NumImportantColors 0
 Image information for oldman.jpg

Filename E:\oldman.jpg
FileModDate 28-Jun-2014 13:25:04
FileSize 43635
Format Jpg
FormatVersion ''
Width 526
Height 526
BitDepth 24
ColorType truecolor
FormatSignature ''
NumberOfSamples 3
CodingMethod Huffman
CodingProcess Progressive
Comment {'*'}

Result:
Images of different types have been displayed and studied.
 2. IMAGE TRANSFORM - FOURIER TRANSFORM
Aim:
The program is to
a) Find the Fourier Transform of an image and also the inverse fourier
transform.
b) Verify the rotation property of Fourier Transform
c) Verify the convolution property of the Fourier Transform

Theory:
The fourier transform, developed by Jean Baptiste Joesph Fourier, is widely
used in the field of image processing. An image is a spatially varying function.
One way to analyse spatial variations is to decompose an image into a set of
orthogonal functions, one such being the fourier functions. A fourier
transform is used to transform an intensity image into the domain of spatial
frequency.
The program first creates an image, finds its fourier transform and displays it.
The inverse fourier transform is found to obtain and display the input image.
The rotation property of Fourier transform states that if an image is rotated
by a certain angle, its fourier transform is also rotated by the same angle. The
program rotates the input image by an angle of 45 degrees. The fourier
transform of the rotated image is found and displayed.
The convolution property of the Fourier transform states that the convolution
of two images in time domain is equivalent to multiplication of the fourier
transforms of the individual images. Two images are created. The convolution
of the two images is carried out and displayed. The fourier transforms of the
individual images are found, multiplied and inverse fourier transform is
applied to the product. We can observe that the output image is the same as
obtained through convolution.
 MATLAB PROGRAM
clc;
clear all;
close all;

% to create an image, find and display its Fourier transform & also to find the
inverse fourier transform %

a=zeros(256,256);
a(110:140,110:140)=1;

subplot(3,1,1), imshow(a), title('Input Image');

a1=log(1+abs(fftshift(fft2(a))));
subplot(3,1,2),imshow(mat2gray(a1)),title('Fourier Transform of the input
image');

a1=fft2(a);
b=ifft2(a1);
subplot(3,1,3), imshow (b), title('Input image obtained by Inverse fourier
transform');

% To verify the rotation property of the Fourier Transform %

a=zeros(256,256);
a(110:140,110:140)=1;
figure;
subplot(2,2,1), imshow(a), title('Input Image');
a1=log(1+abs(fftshift(fft2(a))));
subplot(2,2,2),imshow(mat2gray(a1)),title('Fourier Transform of the input
image');

c=imrotate(a,45,'bilinear','crop');
subplot(2,2,3),imshow(c), title('Input image rotated by 45 degrees');

c1=log(1+abs(fftshift(fft2(c))));
subplot(2,2,4),imshow(mat2gray(c1)), title('Fourier transform of the rotated
image');

% TO VERIFY THE CONVOLUTION PROPERTY %

figure;
a=zeros(256,256);
a(110:140,110:140)=1;
subplot(2,2,1),imshow(a),title('First input image');

b=zeros(256,256);
b(170:200,170:200)=1;
subplot(2,2,2), imshow(b), title('Second input image');

d=conv2(a,b,'same');
subplot(2,2,3), imshow(d),title('Convolution of input images in time domain');

a1=fft2(a);
b1=fft2(b);
e=a1.*b1;
f=fftshift(ifft2(e));
subplot(2,2,4), imshow(f), title({'Result of Multiplication of FFTs'; 'of input
images and IFFT'})

Result:
Thus the fourier transform and inverse fourier transform has been studied as
well as the rotation and convolution properties of the fourier transform.
 3. IMAGE TRANSFORM – DISCRETE COSINE TRANSFORM
Aim:
The program is to find the Discrete Cosine Transform of an image and also the
inverse discrete cosine transform.

Theory:
A discrete cosine transform (DCT) expresses a finite sequence of data points
in terms of a sum of cosine functions oscillating at different frequencies. DCTs
are important to numerous applications in science and engineering, from
lossy compression of audio (e.g. MP3) and images (e.g. JPEG) (where small
high-frequency components can be discarded), to spectral methods for the
numerical solution of partial differential equations. The use of cosine rather
than sine functions is critical for compression, since it turns out that fewer
cosine functions are needed to approximate a typical signal, whereas for
differential equations the cosines express a particular choice of boundary
conditions. In particular, a DCT is a Fourier-related transform similar to the
discrete Fourier transform (DFT), but using only real numbers. The DCTs are
generally related to Fourier Series coefficients of a periodically and
symmetrically extended sequence whereas DFTs are related to Fourier Series
coefficients of a periodically extended sequence. DCTs are equivalent to DFTs
of roughly twice the length, operating on real data with even symmetry (since
the Fourier transform of a real and even function is real and even), whereas in
some variants the input and/or output data are shifted by half a sample. There
are eight standard DCT variants, of which four are common.

Result:
Thus the discrete cosine transform and inverse discrete cosine transform have
been studied.
 MATLAB PROGRAM
clc;
clear all;
close all;

% to create an image, find and display its discrete cosine transform & also to
find the inverse fourier transform %

I4=[1 1 1 1; 1 1 1 1; 1 1 1 1; 1 1 1 1]
D4=dctmtx(4)
output4 = D4*I4*D4
input4=D4'*output4*D4'

% To find the DCT of a given image and the inverse transform %

I = imread('cameraman.tif');
J = dct2(I);
K = idct2(J);
imshow(I),title('Input Image');
figure,imshow(log(abs(J)),[]), colorbar, title('DCT of input image');
figure,imshow(K,[0 255]), title('Input Image obtained by IDCT');
 4. IMAGE TRANSFORM – HADAMARD TRANSFORM
Aim:
To generate Hadamard matrix for N=4, find the Hadamard transform for a
matrix of size 4 x 4. To generate a binary figure of size(256 x 256 and find its
Hadamard transform as well as verify the inverse Hadamard.
Theory:
The Hadamard transform is based on basis functions that are simply +1 or -1,
instead of the more complex sine and cosine functions used in the Fourier
transform. The function Hadamard is used to generate the Hadamard matrix
of required size. The program generates a Hadamard matrix of size 4 x 4. The
Hadamard of the input matrix is found and the inverse also. The program also
generates an image of size 256 x 256. The Hadamard and the inverse
Hadamard are found to verify the program.
 MATLAB PROGRAM

clc;
clear all;
close all;

A=[1 1 1 1; 1 1 1 1; 1 1 1 1; 1 1 1 1];
disp('The input matrix is'),A
H4=hadamard(4);
disp('The hadamard matrix for N=4 is: '), H4
B = 1/4*H4*A*H4;
disp('The hadamard transform of the input matrix is'),B
A=1/4*H4'*B*H4';
disp('The inverse hadamard is'),A

A=zeros(256,256);
A(100:150,100:150)=1;
subplot(1,3,1),imshow(A),title('Input Image');
H256=hadamard(256);
B=1/256*H256*A*H256;
subplot(1,3,2),imshow(B),title('hadamard of the input image');
A=1/256*H256'*B*H256';
subplot(1,3,3),imshow(A),title('Input Image obtained by inverse transform');

Result:
The Hadamard matrix has been generated and verified for an input matrix of
size N=4. The Hadamard of a generated image of size N=256 is generated and
verified.
The input matrix is
A=
1 1 1 1
1 1 1 1
1 1 1 1
1 1 1 1
The hadamard matrix for N=4 is:
H4 =
1 1 1 1
1 -1 1 -1
1 1 -1 -1
1 -1 -1 1
The hadamard transform of the input matrix is
B=
4 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
The inverse hadamard is
A=
1 1 1 1
1 1 1 1
1 1 1 1
1 1 1 1
 5. HISTOGRAM EQUALISATION
Aim:
To study the different methods of histogram equalization and their effects on
enhancing the image. .
Theory:
The histogram of an image with intensity levels in the range [0, L-1] is a
discrete function h(rk) = nk, where rk is the kth intensity level and nk is the
number of pixels in the image with intensity rk. A histogram shows us the
distribution of grey levels in the image. Histograms are the basis for numerous
spatial domain techniques. Histogram manipulation can be used for image
enhancement. In histogram equalisation, the pixels are made to occupy the
entire range of possible intensity levels and in addition the pixels are made to
be distributed uniformly. The net effect will be an image that shows a great
deal of gray level detail and has high dynamic range.
The function imhist is used to display the histogram of a given image. The
figure shows a selection of images and their histograms. Notice the
relationships between the images and their histograms. Note that the high
contrast image has the most evenly spaced histogram.
Function - histeq

g=histeq(f, nlev)

f is the input image array

nlev is the number of intensity levels specified for the output image. If nlev is
equal to the total number of possible levels in the input image then histeq
implements the transformation directly. If nlev is less than the total number of
possible levels in the input image then histeq attempts to distribute the levels
so that they will approximate a flat histogram.
We can see that the histogram equalization done here has produced an image
with washed out appearance.
The reason can be understood by studying the histogram of the equalized
image. The intensity levels have been shifted to the upper one half of the gray
scale thereby giving the image the low contrast washed out appearance.

Function adapthisteq
enhances the contrast of the grayscale image by transforming the values
using contrast-limited adaptive histogram equalization (CLAHE).
CLAHE operates on small regions in the image, called tiles, rather than the
entire image. Each tile's contrast is enhanced. The neighboring tiles are then
combined using bilinear interpolation to eliminate artificially induced
boundaries. The contrast, especially in homogeneous areas, can be limited to
avoid amplifying any noise that might be present in the image.
J = adapthisteq(I,param1,val1,param2,val2...) specifies any of the additional
parameter/value pairs listed in the following table. Parameter names can be
abbreviated, and case does not matter.
 MATLAB PROGRAM
clc;
clear all;
close all;
f=imread('fig0308(a).tif');
imshow(f);
title('Orginal image');
figure, imhist(f)
ylim(‘auto’)
title('Histogram of the original image');
g=histeq(f,256);
figure,imshow(g)
title('Histogram equalised image');
figure,imhist(g)
ylim(‘auto’)
title('Histogram of the equalised image');

clc;
clear all;
close all;
f=imread('fig0310(a).tif');
imshow(f);
title('Orginal image');
figure, imhist(f)
ylim('auto')
title('Histogram of the original image');
g=histeq(f,500);
figure,imshow(g)
title('Histogram equalised image');
figure,imhist(g)
ylim('auto')
title('Histogram of the equalised image');

clc;
clear all;
close all;
f=imread('fig0310(a).tif');
imshow(f);
title('Orginal image');
figure, imhist(f)
ylim('auto')
title('Histogram of the original image');
g=histeq(f,256);
figure,imshow(g)
title('Histogram equalised image');
figure,imhist(g)
ylim('auto')
title('Histogram of the equalised image');
h=adapthisteq(f, 'Numtiles', [25 25], 'cliplimit', 0.05, 'distribution',
'rayleigh');
figure, imshow(h)
title('contrast limited adaptive histogram equalisation');
figure,imhist(h)
title('histogram of contrast limited adaptive histogram equalised image');
 6. IMAGE SMOOTHING
Aim:
To smoothen the given image using different spatial filters and study the
effects.

Theory:
Image smoothing spatial filters are used for blurring and noise reduction. The
output of a smoothing linear filter is simply the average of the pixels
contained in the neighborhood of the filter mask. These filters are also called
as averaging filters. In averaging filters, the value of every pixel in the image is
replaced by the average of the gray levels in the neighborhood defined by the
filter mask. This process results in an image with reduced sharp transitions in
gray levels. However edges which are desirable features of an image are
characterized by sharp transitions in gray levels. Hence averaging filters have
the undesirable effect that blurs the edges. A major use of averaging filters is
in the reduction of irrelevant detail in an image. Irrelevant details refer to the
pixel regions that are small with respect to the size of the filter mask. A spatial
averaging filter in which all coefficients are equal is sometimes is called a box
filter.
We consider an input image of size 500 x 500 pixels as shown. The black
squares at the top are of sizes 3, 5, 9, 15, 25, 35, 45 and 55 pixels. Their
borders are 25 pixels apart. The letters at the bottom range are of sizes 10, 12,
14, 16, 18, 20, 22 and 24 pixels. The large letter is 60 pixels. The vertical bars
are 5 pixels wide, 100 pixels high and separated by 20 pixels. The diameter of
the circles is 25 pixels and the circles are 15 pixels apart. The noisy rectangles
are of size 50 x 120 pixels.
The filter masks used in the program are of sizes 3, 5, 9, 15 and 35 pixels. For
filter mask of size 3 x 3, there is a general slight blurring throughout the
image. But details that are approximately the same size as the filter mask are
affected considerably more. For example the small letter ‘a’ and the fine grain
noise show significant blurring when compared to the rest of the image. The
jagged borders of the characters and gray circles have been smoothed
pleasantly. For the 5 x 5 mask, the result is similar with a slight increase in the
blurring. For n=9, we can see considerably more blurring. For 15 x 15 and 35 x
35 masks, the results are extreme with respect to the size of the objects.
 MATLAB PROGRAM
clc;
clear all;
close all;

f=imread('inputimage.tif');
subplot(1,2,1),imshow(f),title('Input Image');

w=1/9*ones(3);
g1=imfilter(f,w);
subplot(1,2,2),imshow(g1),title('Image smoothened by square averaging filter
mask of size 3 x 3');

w=1/25*ones(5);
g2=imfilter(f,w);
figure, subplot(1,2,1),imshow(g2),title('Image smoothened by square
averaging filter mask of size 5 x 5');

w=1/81*ones(9);
g3=imfilter(f,w);
subplot(1,2,2),imshow(g3),title('Image smoothened by square averaging filter
mask of size 9 x 9');

w=1/225*ones(15);
g4=imfilter(f,w);
figure, subplot(1,2,1),imshow(g4),title('Image smoothened by square
averaging filter mask of size 15 x 15');

w=1/1225*ones(35);
g5=imfilter(f,w);
subplot(1,2,2),imshow(g5),title('Image smoothened by square averaging filter
mask of size 35 x 35');

Result:
Different spatial filters have been used for image smoothening and the effects
have been studied.
 7. IMAGE SHARPENING
Aim:
To sharpen the given image using different spatial filters and study the effects.

Theory:
Sharpening spatial filters seek to highlight fine detail by removing blurring
from images and highlighting the edges. Sharpening filters are based on
spatial differentiation. Differentiation measures the rate of change of a
function. It’s just the difference between subsequent values and measures the
rate of change of the function. The second order derivative simply takes into
account the values both before and after the current value. The 2nd derivative
is more useful for image enhancement than the 1st derivative due to the
stronger response to fine detail and simpler implementation. The first
sharpening filter we will look at is the Laplacian filter, one of the simplest
sharpening filters. Applying the Laplacian to an image we get a new image that
highlights edges and other discontinuities. The result of Laplacian filtering is
not an enhanced image. We have to do more work in order to get our final
image. We have to subtract the Laplacian result from the original image to
generate our final sharpened enhanced image. In the final sharpened image
edges and fine detail are much more obvious. The same process can be
repeated for Sobel and Prewitt filters also.

Result:
Different spatial filters have been used for image sharpening and the effects
have been studied.
 MATLAB PROGRAM
clc;
clear all;
close all;

f=imread('fig0413(a).tif');
subplot(2,2,1),imshow(f),title('input image');

w=1/81*ones(9);
g=imfilter(f,w);
subplot(2,2,2),imshow(g),title('Smoothened Image');

w=fspecial('laplacian',0);
g1=imfilter(g,w,'replicate');
i=f-g1;
subplot(2,2,3),imshow(i);title('special laplacian filter');

w=fspecial('prewitt');
g2=imfilter(g,w);
i=f-g2;
figure, imshow(i);title('Prewitt filter');

w=fspecial('sobel');
g3=imfilter(g,w);
i=f-g3;
figure, imshow(i);title('Sobel filter');
 8. EDGE DETECTION
Aim:
To detect the edges in the given image using different operators.

Theory:
The basic idea behind edge detection is to find places in an image where the
intensity changes rapidly, using one of the following criteria
a) Find places where the first derivative of the intensity is greater in
magnitude than a specified threshold.
b) Find places where the second derivative of the intensity has a zero crossing
The function edge in the MATLAB provides several edge estimations based on
the above criteria. Sobel edge detector computes the gradient by using the
disrete differences between the rows and columns of a discrete neighborhood
where the center pixel is in each row and column is weighed by 2 to provide
smoothing. Prewitt edge detector finds the edges using the prewitts
approximation to the first order derivatives implemented using prewitt mask.
Robert edge detector finds the edges using the Robert approximation to the
first order derivatives implemented using Robert cross gradient mask. The
lapalcian of Gaussian edge detector finds edges by looking for zero crossings
after filtering the input image with a LoG filter. The zero Crossing detector
finds the edges by looking for zero crossings after filtering the input image
with a specified filter. The canny edge detector finds edges by looking for local
maxima of the gradient of the input image. The gradient is calculated using the
derivatives of a Gaussian filter. The method uses two thresholds to detect
strong and weak edges in the output, and includes the weak edges only if they
are connected to the strong edges. Therefore this method is more likely to
detect true weak edges. The program uses all the above edge detectors to find
the edges of a checkerboard image.
 MATLAB PROGRAM
clc;
close all;
clear all;

a = checkerboard(15,5,5)>0.5;
subplot(2,3,1);
imshow(a); title('Input Image');

b = edge(a,'sobel',0.1,'both');
subplot(2,3,2);
imshow(b); title('Sobel Horizontal and vertical Edge Detection ');

b = edge(a,'sobel',0.1,'horizontal');
subplot(2,3,3);
imshow(b); title('Sobel Horizontal Edge Detection');

b = edge(a,'sobel',0.1,'vertical');
subplot(2,3,4);
imshow(b); title('Sobel Vertical Edge Detection');

d = edge(a,'prewitt','horizontal',0.1);
subplot(2,3,5);
imshow(d); title('Prewitt Horizontal Edge Detection');

d = edge(a,'prewitt','vertical',0.1);
subplot(2,3,6);
imshow(d); title('Prewitt Vertical Edge Detection');

d = edge(a,'prewitt','both',0.1);
figure, subplot(2,3,1);
imshow(d); title('Prewitt Horizontal and Vertical Edge Detection');

c = edge(a,'roberts','both');
subplot(2,3,2);
imshow(c); title('Roberts Horizontal and Vertical Edge Detection');
e = edge(a,'canny',.01);
subplot(2,3,3);
imshow(e); title('Canny Edge Detection ');

f = edge(a,'zerocross');
subplot(2,3,4);
imshow(f); title('Zero Cross Edge Detection');

c = edge(a,'log');
subplot(2,3,5);
imshow(c); title('Laplacian of Gaussian Edge Detection');
Result:
The detection of edges in the given image has been carried out using various
edge detectors and verified.
 9. NOISE REMOVAL
Aim:
To remove the noise from the given image.

Theory:
The method of choice for reducing noise is the spatial filtering. Different
spatial filters like the arithmetic mean filter, geometric mean filter, harmonic
mean filter, contraharmonic mean filter, median filter, max filter, min filter,
midpoint filter and alpha trimmed mean filter.
The different types of noises that can occur are random noise, salt and pepper
noise, salt noise, pepper noise, Gaussian noise etc..,
The type of filter used to remove or reduce noise in a given noisy image
depends on the noise present in the image and the image characteristic itself.
In this program we are considering an input image corrupted by 4 different
noises namely salt and pepper noise, salt noise, pepper noise and Gaussian
noise. For each noise we are considering two filters and verify the
performance by calculating the rmse values.
The program can be extended by including all filters for all the types of noises.
 MATLAB PROGRAM
clc;
clear all;
close all;

f=imread('fig0505(a).tif');
g=imnoise(f,'salt & pepper',0.1);
subplot(2,3,1), imshow(g),title('Input image corrupted by salt and pepper
noise');
g1=spfilt(g,'amean',3,3);
subplot(2,3,2),imshow(g1),title('Salt and Pepper Noise removal with
arithmetic mean filter ');
g2=spfilt(g,'median',3,3);
subplot(2,3,3),imshow(g2),title('Salt and Pepper Noise removal with median
filter ');

[M N]=size(f);
R=imnoise2('salt & pepper',M,N,0.1,0);
gp=f;
gp(R==0)=0;
subplot(2,3,4),imshow(gp),title('Image corrupted by Pepper noise');

g3=spfilt(gp,'chmean',3,3,1.5);
subplot(2,3,5),imshow(g3),title('Pepper noise removal by contraharmonic
mean filter');
g4=spfilt(gp,'max',3,3);
subplot(2,3,6),imshow(g4),title('Pepper noise removal by max filter');
[M N]=size(f);
R=imnoise2('salt & pepper',M,N,0,0.1);
gs=f;
gs(R==1)=255;
figure,subplot(2,3,1),imshow(gs),title('Image corrupted by Salt noise');

g5=spfilt(gs,'atrimmed',3,3);
subplot(2,3,2),imshow(g5),title('Salt noise removal by alphatrimmed filter');
g6=spfilt(gs,'min',3,3);
subplot(2,3,3),imshow(g6),title('Salt noise removal by min filter');

gh=imnoise(f,'gaussian');
subplot(2,3,4),imshow(gh),title('Image corrupted by gaussian noise');

g7=spfilt(gh,'gmean',3,3);
subplot(2,3,5),imshow(g7),title('Gaussian noise removal by geometric mean
filter');
g8=spfilt(gh,'hmean',3,3);
subplot(2,3,6),imshow(g8),title('Gaussian removal by harmonic mean filter');

rmse1=compare(f,g);
disp('The RMSE between the input image and the image with salt and pepper
noise'),rmse1
rmse2=compare(f,g1);
disp('The RMSE after applying arithmetic mean filter for reducing salt and
pepper noise'),rmse2
rmse3=compare(f,g2);
disp('The RMSE after applying median filter for reducing salt and pepper
noise'),rmse3
rmse4=compare(f,gp);
disp('The RMSE between the input image and the image with pepper
noise'),rmse4
rmse5=compare(f,g3);
disp('The RMSE after applying contraharmonic mean filter for reducing
pepper noise'),rmse5
rmse6=compare(f,g4);
disp('The RMSE after applying max filter for reducing pepper noise'),rmse6
rmse7=compare(f,gs);
disp('The RMSE between the input image and the image with salt
noise'),rmse7
rmse8=compare(f,g5);
disp('The RMSE after applying alphatrimmed filter for reducing salt
noise'),rmse8
rmse9=compare(f,g6);
disp('The RMSE after applying min filter for reducing salt noise'),rmse9
rmse10=compare(f,gh);
disp('The RMSE between the input image and the image with gaussian
noise'),rmse10
rmse11=compare(f,g7);
disp('The RMSE after applying geometric mean filter for reducing gaussian
noise'),rmse11
rmse12=compare(f,g8);
disp('The RMSE after applying harmonic mean filter for reducing gaussain
noise'),rmse12

Result:
The performance of different filters have been studied for removing noises of
different types.
 10. INVERSE FILTERING
Aim:
To study the effect of inverse filtering in restoring degraded images.

Theory:
In the restoration of images degraded by noise and motion blur, the
simplest approach is to ignore the noise term in the degradation model and
form an estimate of the form F^(u,v) = G(u,v) / H(u,v). The corresponding
estimate of the image is obtained by taking the inverse fourier transform of
F^(u,v) . This approach is called the inverse filtering. Taking noise into
account, we can express our estimate as F^(u,v) = F(u,v) + N(u,v)/H(u,v).
Even if H(u,v) is known exactly, F(u,v) and thereby the input image cannot be
recovered exactly because the noise component is a random function whose
fourier transform N(u,v) is not known. In addition there usually is the
problem of H(u,v) having numerous zeros. Even if the noise term N(u,v) were
negligible, dividing it by vanishing values of H(u,v) would dominate
restoration estimates. The typical approach when attempting inverse filtering
is to form the ratio F^(u,v) = G(u,v) / H(u,v) and then limit the frequency
range for obtaining the inverse, to frequencies near the origin. The idea is that
zeros in H(u,v) are less likely to occur near the origin because the magnitude
of the transform typically is at its highest values in that region. There are
numerous variations of this basic theme.
In the program we first use direct inverse filtering to restore the image
and observe that the process is dominated by noise. Having an estimations of
the noise signal ratio improves the restoration process.

Result:
Image restoration using direct inverse filtering and with the usage of
estimated NSR has been compared.
 MATLAB PROGRAM
clc;
clear all;
close all;

I = im2double(imread('cameraman.tif'));
subplot(2,2,1),imshow(I),title('Input Image');
LEN = 21;
THETA = 11;
PSF = fspecial('motion', LEN, THETA);
blurred = imfilter(I, PSF, 'conv', 'circular');
noise_mean = 0;
noise_var = 0.0001;
blurred_noisy = imnoise(blurred, 'gaussian', ...
noise_mean, noise_var);
subplot(2,2,2),imshow(blurred_noisy),title('Input Image degraded by motion
blur and noise');
estimated_nsr = 0;
wnr2 = deconvwnr(blurred_noisy, PSF, estimated_nsr);
subplot(2,2,3), imshow(wnr2),title('Restoration of Blurred, Noisy Image Using
Ideal Inverse Filter');
estimated_nsr = noise_var / var(I(:));
wnr3 = deconvwnr(blurred_noisy, PSF, estimated_nsr);
subplot(2,2,4),imshow(wnr3), title('Restoration of Blurred, Noisy Image Using
Estimated NSR');
 11. LOSSLESS COMPRESSION
Aim:
To achieve lossless compression of the given image using MATLAB program

Theory:
Image compression addresses the problem of reducing the amount of
data required to represent a digital image. Compression is achieved by
removing one or three data redundancies namely coding redundancy,
spatial/temporal redundancy and irrelevant information.
When coding the gray levels of an image, Huffman codes contain the
smallest possible number of code symbols (bits) per source symbol subject to
the constraint that the source symbols are coded one at a time. Huffman code
generation is not in itself compression. To realize the compression that is built
into a Huffman code, the symbols for which the code was created must be
mapped in accordance with the code generated. The Huffman compression is
achieved by using the function mat2huff. Huffman encoded images are of little
use unless they can be decoded to recreate the original images from which
they were derived. The decoder must compute the Huffman code used to
encode and then inverse map the encoded data to rebuild the input image. The
function huff2mat is used for the purpose.
The lossless predictive coding eliminates the interpixel redundancies of
closely spaced pixels by extracting and coding only the new information in
each pixel. The new information of a pixel is defined as the difference between
the actual and the predicted value of that pixel. The functions mat2lpc and
lpc2mat are used to implement the lossless predictive coding and decoding
processes.
 MATLAB PROGRAM
clc;
clear all;
close all;

f=imread('cameraman.tif');
subplot(1,3,1),imshow(f),title('Input Image');
c=mat2huff(f);
g=huff2mat(c);
cr1=imratio(f,c)
subplot(1,3,2),imshow(g, [0 255]),title('Image Decoded after Huffman
Coding');
rmse=compare(f,g)

c=mat2lpc(f);
cr1=imratio(f,c)
g=lpc2mat(c);
subplot(1,3,3),imshow(g, [0 255]),title('Image Decoded after Lossless
Predictive Coding');
rmse=compare(f,g)

Result:
Lossless compression has been done using Huffman and Lossless Predictive
coding and the results verified.
 12. WAVELET CODING
Aim:
To study the wavelet coding and its applications in image compression

Theory:
The wavelet transform has made it easier to compress, transmit and analyse
many images. Unlike the fourier transform, whose basis functions are
sinusoids, wavelet transforms are based on small waves called wavlets of
varying frequency and limited duration. This allows them to provide the
equivalent of a musical score for an image, revealing not only what notes (or
frequencies) to play but also when to play them. Fourier transforms on the
other hand provide only the notes or frequency information and the temporal
information is lost in the transformation process.
The two dimensional wavelet transform results in a 2-dimensional scaling
function and three directionally sensitive wavelets. The wavelets measure
functional variations and intensity variations for images along different
directions namely horizontal, vertical and diagonal.
The program does single level wavelet decomposition. The inverse wavelet
transform gives the input image. Then the different coefficients namely
approximate coefficients, horizontal detail coefficients and diagonal detail
coefficients are made zero one at a time and inverse transform is applied in
each case to observe the effect. It can be seen that the zeroing of approximate
coefficients results in considerable loss of image information and a large
RMSE value which indicates that the approximate coefficients are very much
essential. The zeroing of horizontal, vertical or diagonal detail coefficients do
not result in much loss of information as can be seen by their lower RMSE
values.
 MATLAB PROGRAM
clc
clear all
close all;

f = imread (‘figure.tif’);
subplot(1,3,1), imshow(f), title(‘Input Image’);

[c,s] = wavefast(f,1,’sym4’);
subplot(1,3,2), wavedisplay(c,s,-6), title(‘One Scael Wavelet Transform’);

f1 = waveback(c,s,’sym4’);
subplot(1,3,3), imshow(mat2gray(f1)), title(‘Image obtained thro inverse
wavelet’);

[nc,y] = wavecut(‘a’c,s);
Figure, subplot(2,2,1), wavedisplay(nc,s,-6), title(‘Approximate coefficients
made zero’);
f1=waveback(nc,s,’sym4);
f1=uint8 (f1)
subplot(2,2,2), imshow(mat2gray(f1)),title(‘Image obtained after zeroing all
approximate coeffiicients’);

[nc,y] = wavecut(‘h’c,s);
figure, subplot(2,2,3), wavedisplay(nc,s,-6), title(‘Horizontal detail coefficients
made zero’);
f2=waveback(nc,s,’sym4);
subplot(2,2,4), imshow(mat2gray(f2)),title(‘Image obtained after zeroing all
horizontal detail coeffiicients’);

[nc,y] = wavecut(‘v’c,s);
figure, subplot(2,2,1), wavedisplay(nc,s,-6), title(‘Vertical detail coefficients
made zero’);
f3=waveback(nc,s,’sym4);
subplot(2,2,2), imshow(mat2gray(f3)),title(‘Image obtained after zeroing all
vertical detail coeffiicients’);
[nc,y] = wavecut(‘d’c,s);
figure, subplot(2,2,3), wavedisplay(nc,s,-6), title(‘Diagonal detail coefficients
made zero’);
f4=waveback(nc,s,’sym4);
subplot(2,2,4), imshow(mat2gray(f4)),title(‘Image obtained after zeroing all
diagonal detail coeffiicients’);
rmse1=compare(f,f1);
rmse2=compare(f,f2);
rmse3=compare(f,f3);
rmse4=compare(f,f4);