display image similarity using pie and bar chart

Hi guys, i want to know whether it is possible to display the similarities between 2 or more images using pie chart and bar chart in percentages.

Answer 1

[Subhadeep Koley]

I have used the ssim function to measure the similarity between to images. ssim metric value lies in the range of [0, 1]. Higher ssim value represents higher similarity. Therefore, I have used 1-ssimVal to caculate the dissimilarity. Refer the code below,

close all; clc;

% Read the original image
img = imread('cameraman.tif');

% Adding noise in the original image for demo
noisyImg = imnoise(img, 'gaussian', 0, 0.05);

% Calculate Structural Similarity Index (SSIM)
ssimVal = ssim(img,noisyImg);

% (1-ssimVal) will give the dissimilarity
values = [ssimVal, (1-ssimVal)];

% Display results
figure('units', 'normalized', 'outerposition', [0 0 1 1])
subplot(2, 2, 1)
imshow(img)
title('Original image')

subplot(2, 2, 2)
imshow(noisyImg)
title('Noisy image')

subplot(2, 2, 3)
pie(values)
labels = {'Similarity(%)', 'Dissimilarity(%)'};
legend(labels, 'Location', 'bestoutside')
title('Similarity pie chart')

subplot(2, 2, 4)
xValues = categorical({'Similarity(%)', 'Dissimilarity(%)'});
yValues = [ssimVal (1-ssimVal)].*100;
han = bar(xValues, yValues);
xtips1 = han(1).XEndPoints;
ytips1 = han(1).YEndPoints;
labels1 = string(han(1).YData);
text(xtips1, ytips1, labels1, 'HorizontalAlignment', 'center',...
    'VerticalAlignment', 'bottom')
title('Similarity bar chart')

Hope this helps!

[Subhadeep Koley]

@ nat larsey It is not clear what you exactly meant by 'compare'. SSIM compare how visually similar two images are. You want something like Cross Correlation metric between those images?

[nat larsey]

yes sir