I tried it but it didnt work

Check your current format and confirm it to be 'long' (as suggested by Khademi). I don think there should be any other problem as Matlab uses double as its default data type. Check whether you or your function uses - floor or fix functions. As these are used to round off and to separate the integer and fractions.

How to check for floor and fix functions

Can you please be more specific in your questions. So that someone could help out. What Matlab function you are using ? or have you written function to find MSE ? If so post it.

this is my code which i am trying

format long

I = imread('building128_128.tif');

I1 = imnoise(I,'gaussian');

f = [];

f1 = [];

MSE = 0;

a = 0;

[m,n] = size(I);

f = I;

[m1,n1] = size(I1);

%imshow(I1);

f1 = I1;

for i =1:m

for j=1:n

a = (f(i,j)-f1(i,j))*(f(i,j)-f1(i,j));

MSE = MSE + a;

end

end

fprintf('%f',MSE);

% MSE = MSE/(m*n);

% fprintf('%f',MSE);

%PSNR = 10 * log10((255*255)/(MSE));

building is an 128*128 image

Remove the comments part while trying the program

Check the input image class I. It would be of the class ( - unsigned integer) which other than double and hence the MSE value is 0 even it is 0.0089.

Convert the input image I to be of class double. It would solve your problem.

@ sriram ramakrishnan i would try and tell it you

I used im2double function and the code worked

thks to both sriram and april khademi