Scribd
Upload
54 views

Question-1 Code:: Name - Bhumika Verma Reg. No. - 19BCE1418 Teacher: Dr. S. Geetha Subject: CBIR LAB (L45+L46)

The document contains code for an image retrieval lab experiment. The code takes in an image dataset, extracts visual features from the images using an auto-correlogram method, calculates distances between features of images and a query image, and returns the top similar images to the query. It displays the query image and top similar images found. It also prints the total number of features extracted and time taken for the process.

Uploaded by

bhumika.verma00
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
54 views

Question-1 Code:: Name - Bhumika Verma Reg. No. - 19BCE1418 Teacher: Dr. S. Geetha Subject: CBIR LAB (L45+L46)

The document contains code for an image retrieval lab experiment. The code takes in an image dataset, extracts visual features from the images using an auto-correlogram method, calculates distances between features of images and a query image, and returns the top similar images to the query. It displays the query image and top similar images found. It also prints the total number of features extracted and time taken for the process.

Uploaded by

bhumika.verma00
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 12

NAME – BHUMIKA VERMA

Reg. No. – 19BCE1418

Teacher: Dr. S. Geetha

Subject: CBIR LAB (L45+L46)

LAB-5

QUESTION-1
CODE:
% Variables

resultFile = 'rgb_image_features.xlsx';

total_images = 20;

levels = 256;

distance_vector = [1 3];

total_features = 2*levels*size(distance_vector,2);

dataset = zeros(total_images, total_features+1);

queryImgFileName = '11.jpg';

queryImgFeatures = imgFeatures(queryImgFileName,distance_vector,levels);

CSDwithQueryImg = zeros(total_images, 2);

total_solutions = 6;

% Generate Labels/Column headers for the dataset

labels = strings(1, total_features + 1);


labels(1,1) = "Image Name";

for d = 1:size(distance_vector, 2)

offset = 256*2*(d-1);

for bin = 0:255

h_idx = offset + 0*256 + (bin + 1) + 1;

labels(1, h_idx) = sprintf("H-%d-%d,%d", d, bin, bin);

v_idx = offset + 1*256 + (bin + 1) + 1;

labels(1, v_idx) = sprintf("V-%d-%d,%d", d, bin, bin);

end

end

writematrix(labels, resultFile);

% Retrieve the features for all images

for idx = 0:total_images-1

imgFileName = sprintf("%d.jpg", idx);

features = imgFeatures(imgFileName, distance_vector, levels);

imgDetails = [imgFileName string(features)];

% Store the features in a matrix

dataset(idx+1, 1:total_features+1) = [idx features];

% Store the features in an Excel file

% writematrix(imgDetails, resultFile, 'WriteMode','append');

end

for idx = 0:total_images-1

% Retreive features for the test img


testImgFeatures = dataset(idx+1, 2:total_features+1);

% Calculate the Chi-Squared Distance b/w the features

CSD_raw = ((queryImgFeatures - testImgFeatures).^2)./queryImgFeatures;

% The above matrix can have NaN due to zero counts in queryImgFeatures

% So, used rmmissing to clean those missing values

CSD = sum(rmmissing(CSD_raw));

% Store the data in a string matrix

CSDwithQueryImg(idx+1,:) = [idx CSD];

end

% Sort the images w.r.t. to Euclidean Distance

CSDwithQueryImg = sortrows(CSDwithQueryImg, 2);

% Display the query image

subplot(3, 3, 2);

imshow(getImgFilePath(queryImgFileName));

title('Query Image');

% Display similar images

for idx = 1:total_solutions

subplot(3, 3, idx + 3);

imshow(getImgFilePath(sprintf('%d.jpg',CSDwithQueryImg(idx, 1))));

title(sprintf('Similar Image %d', idx));

end

% Function to retrieve file path


function filePath = getImgFilePath(imgName)

filePath = sprintf('folder1/%s', imgName);

end

% Function to extract features of the given image

function features = imgFeatures(fileName, distance_vector, levels)

% Default parameters for levels

if nargin == 2

levels = 256;

end

filePath = getImgFilePath(fileName);

img = imread(filePath);

% Convert image to grayscale

img = im2gray(img);

% Use custom auto-correlogram function: acg

[hc, vc] = acg(img, distance_vector, levels);

% Fill data back into the return parameter

features = zeros(1, 2*levels*size(distance_vector,2));

for d = 1:size(distance_vector, 2)

start_idx = (d-1) * 2 * levels + 1;

end_idx = (d) * 2 * levels;

features(1, start_idx:end_idx) = [hc(:,d).' vc(:,d).'];


end

end

function [horizontal_count, vertical_count] = acg(img, distances, levels)

% Check if levels provided or not

if nargin == 2

levels = 256;

end

[Y, X] = size(img);

% Image quantization

img = gray2ind(img, levels);

% Set variable sizes

[~, num_of_distances] = size(distances);

horizontal_count = zeros(levels, num_of_distances);

vertical_count = zeros(levels, num_of_distances);

% For each row

for r = 1:Y

% For each column

for c = 1:X

% For each distance

for d = 1:num_of_distances

D = distances(d);

value = img(r,c); % Get the value


% Increment the resp. counter, if pixels equivalent

if(r + D <= Y && img(r + D, c) == value)

horizontal_count(value+1, d) = horizontal_count(value+1, d) + 1;

end

if(c + D <= X && img(r, c + D) == value)

vertical_count(value+1, d) = vertical_count(value+1, d) + 1;

end

end

end

end

end

OUTPUT:
QUESTION 2:
CODE :
% Start stopwatch
tic
% Variables
resultFile = 'rgb_image_features.xlsx';
total_images = 20;
levels = 256;
distance_vector = [1 3];
total_features = 2*levels*size(distance_vector,2);
dataset = zeros(total_images, total_features+1);
queryImgFileName = '13.jpg';
queryImgFeatures = imgFeatures(queryImgFileName,distance_vector,levels);
CSDwithQueryImg = zeros(total_images, 2);
total_solutions = 6;

% Generate Labels/Column headers for the dataset


labels = strings(1, total_features + 1);
labels(1,1) = "Image Name";
for d = 1:size(distance_vector, 2)
offset = 256*2*(d-1);
for bin = 0:255
h_idx = offset + 0*256 + (bin + 1) + 1;
labels(1, h_idx) = sprintf("H-%d-%d,%d", d, bin, bin);
v_idx = offset + 1*256 + (bin + 1) + 1;
labels(1, v_idx) = sprintf("V-%d-%d,%d", d, bin, bin);
end
end
writematrix(labels, resultFile);

% Retrieve the features for all images


for idx = 0:total_images-1
imgFileName = sprintf("%d.jpg", idx);
features = imgFeatures(imgFileName, distance_vector, levels);
imgDetails = [imgFileName string(features)];
% Store the features in a matrix
dataset(idx+1, 1:total_features+1) = [idx features];
% Store the features in an Excel file
% writematrix(imgDetails, resultFile, 'WriteMode','append');
end
for idx = 0:total_images-1
% Retreive features for the test img
testImgFeatures = dataset(idx+1, 2:total_features+1);
% Calculate the Chi-Squared Distance b/w the features
CSD_raw = ((queryImgFeatures - testImgFeatures).^2)./queryImgFeatures;
% The above matrix can have NaN due to zero counts in queryImgFeatures
% So, used rmmissing to clean those missing values
CSD = sum(rmmissing(CSD_raw));
% Store the data in a string matrix
CSDwithQueryImg(idx+1,:) = [idx CSD];
end

% Sort the images w.r.t. to Euclidean Distance


CSDwithQueryImg = sortrows(CSDwithQueryImg, 2);

% Display the query image


subplot(3, 3, 2);
imshow(getImgFilePath(queryImgFileName));
title('Query Image');
% Display similar images
for idx = 1:total_solutions
subplot(3, 3, idx + 3);
imshow(getImgFilePath(sprintf('%d.jpg',CSDwithQueryImg(idx, 1))));
title(sprintf('Similar Image %d', idx));
end

% Display the total features & time elapsed


fprintf("Total features: %d\n", total_features);
% Stop stopwatch
toc

% Function to retrieve file path


function filePath = getImgFilePath(imgName)
filePath = sprintf('folder1/%s', imgName);
end

% Function to extract features of the given image


function features = imgFeatures(fileName, distance_vector, levels)
if nargin == 2
levels = 256;
end
filePath = getImgFilePath(fileName);

img = imread(filePath);
% Convert image to grayscale
% img = im2gray(img);
img = img(:,:,1);
% Use custom auto-correlogram function: acg
[hc, vc] = acg(img, distance_vector, levels);

% Fill data back into the return parameter


features = zeros(1, 2*levels*size(distance_vector,2));
for d = 1:size(distance_vector, 2)
start_idx = (d-1) * 2 * levels + 1;
end_idx = (d) * 2 * levels;
features(1, start_idx:end_idx) = [hc(:,d).' vc(:,d).'];
end
end
function [horizontal_count, vertical_count] = acg(img, distances, levels)
% Check if levels provided or not
if nargin == 2
levels = 256;
end

[Y, X] = size(img);
% Image quantization
img = gray2ind(img, levels);

% Set variable sizes


[~, num_of_distances] = size(distances);
horizontal_count = zeros(levels, num_of_distances);
vertical_count = zeros(levels, num_of_distances);

% For each row


for r = 1:Y
% For each column
for c = 1:X
% For each distance
for d = 1:num_of_distances
D = distances(d);
value = img(r,c); % Get the value
% Increment the resp. counter, if pixels equivalent
if(r + D <= Y && img(r + D, c) == value)
horizontal_count(value+1, d) = horizontal_count(value+1, d) + 1;
end
if(c + D <= X && img(r, c + D) == value)
vertical_count(value+1, d) = vertical_count(value+1, d) + 1;
end
end
end
end
end
OUTPUT –
1)

2)

3)
4)
5)

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy