| |
| 20,777 bytes | 10,816 bytes |
The above images differ by the quality of the JPEG compression.
» im1 = imread('fashion.jpg');
» im2 = imread('fashion2.jpg');
» im1 = im2double(im1);
» im2 = im2double(im2);
» im1g = rgb2gray(im1);
» im2g = rgb2gray(im2);
» imwrite(im1g,'im1g.jpg');
» imwrite(im2g,'im2g.jpg');
» diff = im1g - im2g;
» std2(diff)
ans =
0.0256
» mean2(diff.*diff)
ans =
6.5492e-004
» sqrt(ans)
ans =
0.0256
» mean2(diff)
ans =
2.1021e-004
» diff1 = 0.5*diff+0.5;
» imshow(diff1);
» imwrite(diff1,'diff1.jpg');
» imhist(diff1);
» cmax = max(max(diff))
cmax =
0.2135
» cmin = min(min(diff))
cmin =
-0.1777
» diff2 = 0.5*diff/max(cmax,-cmin)+0.5;
» imwrite(diff2,'diff2.jpg');
» imhist(diff2);
» diff3 = (diff-cmin)/(cmax-cmin); » imwrite(diff3,'diff3.jpg');
» imhist(diff3)
» mse = mean2(diff.*diff)
mse =
6.5492e-004
» rmse = sqrt(mse)
rmse =
0.0256
» imstats(diff)
Image has 51 of 256 possible levels
zmax zmin range contrast
0.2135 -0.1777 0.3911 1.832
mean std skewness kurtosis
0.0002102 0.02559 0.05473 5.1
The skewness and kurtosis are not affected by a change in scale.
» imstats(diff2)
Image has 203 of 256 possible levels
zmax zmin range contrast
1 0.08383 0.9162 0.9162
mean std skewness kurtosis
0.5005 0.05994 0.05473 5.1
function imstats(im)
%
% Calculate image statistics
%
[counts, x] = imhist(im);
N = sum(counts);
f = find(counts);
n = length(f);
nlevels = length(x);
if (n <nlevels)
disp(sprintf('Image has %d of %d possible levels\n',n, nlevels));
end
z = im2double(im);
% contrast
zmax = max(max(z));
zmin = min(min(z));
range = zmax-zmin;
modulation = range/max(abs([zmin zmax]));
disp(sprintf(' zmax\t zmin\t range\t contrast'));
disp(sprintf('%8.4g\t%8.4g\t%8.4g\t%10.4g\n',zmax,zmin,range,modulation));
% moments
avg = mean2(z);
z = z - avg;
sig = sqrt(mean2(z.*z));
z = z/sig;
skewness = mean2(z.^3);
kurtosis = mean2(z.^4)-3;
disp(sprintf(' mean\t std\t skewness\t kurtosis'));
disp(sprintf('%8.4g\t%8.4g\t%10.4g\t%10.4g',avg,sig,skewness,kurtosis));
Maintained by John Loomis, last updated 15 Jan 2001