Calculating UACI, NPCR for 2 images

Presently I have a problem in implementing the UACI and PSNR code for 2 images which are respectively Original & Cipher Images. I need the code for calculating it......

What would the Code? I don't have much idea about it.. Do I need to find UACI between 2 images or only 1 image will be sufficient

Answer 1

[Image Analyst]

See my demo:

% Demo to calculate PSNR of a gray scale image.
% http://en.wikipedia.org/wiki/PSNR
% by ImageAnalyst 
clc; 
close all; 
clearvars; 
workspace;
% Read in standard MATLAB demo image.
grayImage = imread('cameraman.tif');
[rows columns] = size(grayImage);
subplot(2, 2, 1);
imshow(grayImage, []);
title('Original Grey Scale Image');
set(gcf, 'Position', get(0,'Screensize')); % Maximize figure.
% Add noise to it.
noisyImage = imnoise(grayImage, 'gaussian', 0, 0.003);
subplot(2, 2, 2);
imshow(noisyImage, []);
title('Noisy Image');

% Calculate mean square error.
mseImage = (double(grayImage) - double(noisyImage)) .^ 2;
subplot(2, 2, 3);
imshow(mseImage, []);
title('MSE Image');
mse = sum(sum(mseImage)) / (rows * columns);
% Calculate PSNR (Peak Signal to noise ratio).
PSNR = 10 * log10( 256^2 / mse);
message = sprintf('The mean square error is %.2f\nThe PSNR = %.2f', ...
  mse, PSNR);
msgbox(message);
% set(gcf, 'Position', get(0,'Screensize')); % Maximize figure.

After this example you should be able to use similar code to easily calculate NPCR and UACI, according to http://www.cyberjournals.com/Papers/Apr2011/05.pdf. You need both images to calculate them since they are a measure of the differences between the two images.

[Komeil]

The NPCR is different from the PSNR.

[Walter Roberson]

Image Analyst says "After this example you should be able to use similar code to easily calculate NPCR" -- in other words, he has not given the code here, but he has given a framework that can be adapted to calculate NPCR by anyone who looks up the formula for NPCR.

Answer 2

[Hyderkkk]

How do i get the NPCR and UACI of an original speech and encryption in matlab code ?

[Image Analyst]

Check the File Exchange: NPCR and UACI measurements with statistical tests

[Hyderkkk]

This file give the relation between two encrypted images of same size and type but I need the relation between original speech and encrypted speech . thank you for answer

[Image Analyst]

You're welcome, but that's all I can contribute. If that doesn't work for you, you'll have to start writing code.

[merr LAMINe]

if you use MATLAB , the samples of the speech normalized in the interval [-1,1], so as a personal idea, I think it calculated with this code (Rmrq : uaci is calculated between 2 encrypted speech )

function [ uc_value1 ] = uaci( s1,s2 )

l=length(s1);
sm=sum(abs(s1-s2));
uc_value1=(100/l)*(sm/2);%2 instead of 65535 (1-(-1)=2)

end

[Kiki Kiosk]

Hey, can you help me?

I must calculate uaci between two audio files, but it's not working right ..

This is my code

uaci_score_mine = sum(abs(data(:) - data1(:))) / 2/ N*100;

I have tried with

uaci_score_mine = normcdf(sum(abs(data(:) - data1(:))) / 2/ N)*100;

and

uaci_score_mine = normcdf(sum(abs(data(:) - data1(:))) / 65535/ N)*100;

and

uaci_score_mine = normcdf(sum(abs(data(:) - data1(:)) / 65535)/ N)*100;

but the results are still not that i expect ..

data and data1 are the audio files

N - length(data);

10x for your attention

[Walter Roberson]

Not clear to me where the 65535 is coming from? If you had used 65536 I might have thought it was the same as the 256^2 PSNR formula that Image Analyst used ?

[Kiki Kiosk]

This is the formula I used to calculate.

[Walter Roberson]

are you reading the audio with the 'native' option?

[Kiki Kiosk]

this is my code:

[data,fs]=audioread('Sample1.wav','native');
[data1,fs2]=audioread('SampleCrypt.wav','native');
 N = length(data);
uaci_score_mine = normcdf(sum(abs(data(:) - data1(:))) / 2/ N)*100; 
fprintf('\n The uaci value is %0.15f \n', uaci_score_mine);

[Samaa Yasser]

@Image Analyst hey, please i want to ask question, i want to know why should the value of uaci in image encryption is about 33. ?

[Image Analyst]

I have no idea. Who said uaci should always be about 33?

[Walter Roberson]

This file give the relation between two encrypted images of same size and type but I need the relation between original speech and encrypted speech

Are you familiar with the encryption technique that is often called ROT13? It is a caeser cypher in which each letter in the first half of the alphabet is replaced with the letter in the corresponding position in the second half of the alphabet, and vice versa, so 'A' gets replaced with the character 13 further on (which is 'N'), and 'p' would get replaced with the character 13 earlier (which is 'c')

Now take the original image, and apply ROT13 to it. Then take the result, and apply ROT13 to that, for even stronger encryption.

Now take the relationship between that double-ROT13 encrypted image and the encryption you found using a different method. As they are both encrypted image, there should be no problem comparing the two using an algorithm that compares encrypted images, right?

... But it turns out that applying ROT13 twice always gets you back exactly the original data. Therefore the original data is itself an encrypted version of the original data... just encrypted with a very weak encryption. And so you can use your algorithm that compares two encrypted images in order to compare the original image to the version of it you encrypted some other way than double-ROT13.