Assignment 2:

Digital Image Processing

Submitted by

Kanish Mahato
21103070
CSE

1/11
1. Read 2 different images and add the following noises in the image.
(a) Gaussian Noise
(b) Erlang Noise
(c) Exponential Noise
(d) Uniform Noise
(e) Salt and Pepper Noise

Code :-

img1 = imread('image1.tiff');
img2 = imread('image2.tiff');

gaussianimg1 = imnoise(img1, 'gaussian');

gaussianimg2 = imnoise(img2, 'gaussian');

shape = 2;
scale = 1;
erlangimg1 = img1 + uint8(gamrnd(shape, scale, size(img1)) * 255);
erlangimg2 = img2 + uint8(gamrnd(shape, scale, size(img2)) * 255);

lambda = 1;
expimg1 = exprnd(1/lambda, size(img1));
expimg2 = exprnd(1/lambda, size(img2));
expimg1 = uint8(double(img1) + expimg1 * 255);
expimg2 = uint8(double(img2) + expimg2 * 255);

uniformimg1 = imnoise(img1, 'salt & pepper', 0.1);

uniformimg2 = imnoise(img2, 'salt & pepper', 0.1);

saltPepperimg1 = imnoise(img1, 'salt & pepper');

saltPepperimg2 = imnoise(img2, 'salt & pepper');

figure;
subplot(3, 4, 1), imshow(img1), title('Original Image 1');
subplot(3, 4, 2), imshow(img2), title('Original Image 2');
subplot(3, 4, 3), imshow(gaussianimg1), title('Gaussian Noise 1');
subplot(3, 4, 4), imshow(gaussianimg2), title('Gaussian Noise 2');

2/11
subplot(3, 4, 5), imshow(erlangimg1), title('Erlang Noise 1');
subplot(3, 4, 6), imshow(erlangimg2), title('Erlang Noise 2');

subplot(3, 4, 7), imshow(expimg1), title('Exponential Noise 1');

subplot(3, 4, 8), imshow(expimg2), title('Exponential Noise 2');

subplot(3, 4, 9), imshow(uniformimg1), title('Uniform Noise 1');

subplot(3, 4, 10), imshow(uniformimg2), title('Uniform Noise 2');

subplot(3, 4, 11), imshow(saltPepperimg1), title('Salt & Pepper Noise

1');
subplot(3, 4, 12), imshow(saltPepperimg2), title('Salt & Pepper Noise
2');

Output :-

3/11
2. Remove the noise from the noisy images obtained in Ques 1
using the following filters:
(a) Arithmetic mean filter
(b) Geometric mean filter
(c) Contraharmonic filter
(d) Median Filter
(e) Weiner Filter

Code :-
Contraharmonic filter code as separate file:-
function output = contraharmonicFilter(img, m, n, Q)

paddedImage = padarray(double(img), [floor(m/2), floor(n/2)],

'symmetric');
[rows, cols] = size(img);
output = zeros(rows, cols);

for i = 1:rows
for j = 1:cols
region = paddedImage(i:i+m-1, j:j+n-1);
numerator = sum(sum(region.^(Q + 1)));
denominator = sum(sum(region.^Q));
output(i,j) = numerator / denominator;
end
end

output = uint8(output);
end

img1 = imread('skull.jpg');

gaussianimg1 = imnoise(img1, 'gaussian');

h = fspecial('average', [3 3]);
arithmeticmeanfilteredimg_gaussian = imfilter(gaussianimg1, h);
geometricmeanfilteredimg_gaussian =
exp(imfilter(log(double(gaussianimg1) + 1), h, 'replicate')) - 1;
geometricmeanfilteredimg_gaussian =
uint8(geometricmeanfilteredimg_gaussian);
q = 1.5;
contraharmonicfilteredimg_gaussian =
contraharmonicFilter(gaussianimg1, 3, 3, q);
medianfilteredimg_gaussian = medfilt2(gaussianimg1, [3 3]);

4/11
weinerfilteredimg_gaussian = wiener2(gaussianimg1, [5 5]);

lambda = 1;
erlang_noise = -1/lambda * log(rand(size(img1)));
erlangimg1 = uint8(double(img1) + erlang_noise);
arithmeticmeanfilteredimg_erlang = imfilter(erlangimg1, h);
geometricmeanfilteredimg_erlang =
exp(imfilter(log(double(erlangimg1) + 1), h, 'replicate')) - 1;
geometricmeanfilteredimg_erlang =
uint8(geometricmeanfilteredimg_erlang);
contraharmonicfilteredimg_erlang =
contraharmonicFilter(erlangimg1, 3, 3, q);
medianfilteredimg_erlang = medfilt2(erlangimg1, [3 3]);
weinerfilteredimg_erlang = wiener2(erlangimg1, [5 5]);

exp_noise = -1/lambda * log(rand(size(img1)));

exponentialimg1 = uint8(double(img1) + exp_noise);
arithmeticmeanfilteredimg_exp = imfilter(exponentialimg1, h);
geometricmeanfilteredimg_exp =
exp(imfilter(log(double(exponentialimg1) + 1), h, 'replicate')) -
1;
geometricmeanfilteredimg_exp =
uint8(geometricmeanfilteredimg_exp);
contraharmonicfilteredimg_exp =
contraharmonicFilter(exponentialimg1, 3, 3, q);
medianfilteredimg_exp = medfilt2(exponentialimg1, [3 3]);
weinerfilteredimg_exp = wiener2(exponentialimg1, [5 5]);

uniform_noise = rand(size(img1)) * 0.1;

uniformimg1 = uint8(double(img1) + uniform_noise);
arithmeticmeanfilteredimg_uniform = imfilter(uniformimg1, h);
geometricmeanfilteredimg_uniform =
exp(imfilter(log(double(uniformimg1) + 1), h, 'replicate')) - 1;
geometricmeanfilteredimg_uniform =
uint8(geometricmeanfilteredimg_uniform);
contraharmonicfilteredimg_uniform =
contraharmonicFilter(uniformimg1, 3, 3, q);
medianfilteredimg_uniform = medfilt2(uniformimg1, [3 3]);
weinerfilteredimg_uniform = wiener2(uniformimg1, [5 5]);

saltpepperimg1 = imnoise(img1, 'salt & pepper', 0.02);

arithmeticmeanfilteredimg_sp = imfilter(saltpepperimg1, h);

5/11
geometricmeanfilteredimg_sp =
exp(imfilter(log(double(saltpepperimg1) + 1), h, 'replicate')) -
1;
geometricmeanfilteredimg_sp = uint8(geometricmeanfilteredimg_sp);
contraharmonicfilteredimg_sp =
contraharmonicFilter(saltpepperimg1, 3, 3, q);
medianfilteredimg_sp = medfilt2(saltpepperimg1, [3 3]);
weinerfilteredimg_sp = wiener2(saltpepperimg1, [5 5]);

figure;
subplot(5, 3, 1), imshow(gaussianimg1), title('Gaussian Noise');
subplot(5, 3, 2), imshow(arithmeticmeanfilteredimg_gaussian),
title('AM Filter (Gaussian)');
subplot(5, 3, 3), imshow(geometricmeanfilteredimg_gaussian),
title('GM Filter (Gaussian)');
subplot(5, 3, 4), imshow(erlangimg1), title('Erlang Noise');
subplot(5, 3, 5), imshow(arithmeticmeanfilteredimg_erlang),
title('AM Filter (Erlang)');
subplot(5, 3, 6), imshow(geometricmeanfilteredimg_erlang),
title('GM Filter (Erlang)');
subplot(5, 3, 7), imshow(exponentialimg1), title('Exponential
Noise');
subplot(5, 3, 8), imshow(arithmeticmeanfilteredimg_exp),
title('AM Filter (Exponential)');
subplot(5, 3, 9), imshow(geometricmeanfilteredimg_exp), title('GM
Filter (Exponential)');
subplot(5, 3, 10), imshow(uniformimg1), title('Uniform Noise');
subplot(5, 3, 11), imshow(arithmeticmeanfilteredimg_uniform),
title('AM Filter (Uniform)');
subplot(5, 3, 12), imshow(geometricmeanfilteredimg_uniform),
title('GM Filter (Uniform)');
subplot(5, 3, 13), imshow(saltpepperimg1), title('Salt & Pepper
Noise');
subplot(5, 3, 14), imshow(arithmeticmeanfilteredimg_sp),
title('AM Filter (Salt & Pepper)');
subplot(5, 3, 15), imshow(medianfilteredimg_sp), title('Median
Filter (Salt & Pepper)');

figure;
subplot(5, 2, 1), imshow(contraharmonicfilteredimg_gaussian),
title('Contraharmonic (Gaussian)');
subplot(5, 2, 2), imshow(weinerfilteredimg_gaussian),
title('Wiener Filter (Gaussian)');

6/11
subplot(5, 2, 3), imshow(contraharmonicfilteredimg_erlang),
title('Contraharmonic (Erlang)');
subplot(5, 2, 4), imshow(weinerfilteredimg_erlang), title('Wiener
Filter (Erlang)');
subplot(5, 2, 5), imshow(contraharmonicfilteredimg_exp),
title('Contraharmonic (Exponential)');
subplot(5, 2, 6), imshow(weinerfilteredimg_exp), title('Wiener
Filter (Exponential)');
subplot(5, 2, 7), imshow(contraharmonicfilteredimg_uniform),
title('Contraharmonic (Uniform)');
subplot(5, 2, 8), imshow(weinerfilteredimg_uniform),
title('Wiener Filter (Uniform)');
subplot(5, 2, 9), imshow(contraharmonicfilteredimg_sp),
title('Contraharmonic (Salt & Pepper)');
subplot(5, 2, 10), imshow(weinerfilteredimg_sp), title('Wiener
Filter (Salt & Pepper)');

Output :-

7/11
3. Read a color image in RGB format and display it.
Change the color space of the image to the following:
(a) CMY
(b) HIS
(c) Lab

Code :-
img_rgb = imread('image1.tiff');
figure;

8/11
subplot(2, 2, 1), imshow(img_rgb), title('RGB Image');

img_cmy = 255 - img_rgb;

subplot(2, 2, 2), imshow(img_cmy), title('CMY Image');

img_hsv = rgb2hsv(img_rgb);
subplot(2, 2, 3), imshow(img_hsv), title('HSV Image');

img_lab = rgb2lab(img_rgb);
subplot(2, 2, 4), imshow(lab2rgb(img_lab)), title('Lab Image');

Output :-

4.Read a color image and smooth it using the following

filters:
Averaging filter
Gaussian filter

9/11
Code :-
img_rgb = imread('image1.tiff');
figure;
subplot(1, 3, 1), imshow(img_rgb), title('Original Image');

h_avg = fspecial('average', [3 3]);

img_avg = imfilter(img_rgb, h_avg);
subplot(1, 3, 2), imshow(img_avg), title('Averaging Filter');

h_gaussian = fspecial('gaussian', [3 3], 0.5);

img_gaussian = imfilter(img_rgb, h_gaussian);
subplot(1, 3, 3), imshow(img_gaussian), title('Gaussian Filter');

Output :-

5. Read a color image and sharpen it using Laplacian filter

10/11
Code:-
img_rgb = imread('image1.tiff');
figure;
subplot(1, 3, 1), imshow(img_rgb), title('Original Image');

h_laplacian = fspecial('laplacian', 0.2);

img_laplacian = imfilter(img_rgb, h_laplacian);
subplot(1, 3, 2), imshow(img_laplacian), title('Laplacian Image');

img_sharpened = uint8(double(img_rgb) - double(img_laplacian));

subplot(1, 3, 3), imshow(img_sharpened), title('Sharpened Image');
Output :-

11/11