how can i get the pixel values of L , a, b in Lab space?

i write this code data(m,n).color= impixel(imlab,m,n); imlab is an image in lab space and the result of running the code has 3 number i dont know is it correct? if yes which one is a ? and how can i use this value to calculate the difference between 2 pixels color? and do you suggest using DELTA E?

Answer 1

[Image Analyst]

L is plane 1, a is 2, and b is 3.

labImage = rgb2lab(rgbImage);
lImage = labImage(:, :, 1);
aImage = labImage(:, :, 2);
bImage = labImage(:, :, 3);
imshow(labImage);
hp = impixelinfo();

Yes you should use delta E to determine color difference.

deltaEImage = sqrt((lImage - lRef) .^ 2 + ...
                   (aImage - aRef) .^ 2 + ...
                   (bImage - bRef) .^ 2);
imshow(deltaEImage, []);
hp = impixelinfo();

Note, these are not TRUE CIELAB values like you'd get from a spectrophotometer and simply use the "book formula". If you change the exposure of your camera, you will get new RGB values, and hence new LAB values even though your sample did not change at all. To avoid that, you'd need to take the much more complicated step of doing a color calibration of your system with known standards like the X-Rite Color Checker Chart.

[mandana]

thanks so much. but the Delta code that you have posted is correct and complete? or i should search the correct Delta code?

[Image Analyst]

Which delta E do you want? That is the CIE 1976 delta E - the simplest. There are better but more complicated versions, like DeltaE2000. Because they're so complicated I don't advise using them unless you're also using calibrated intensity - it's overkill for no reason. No need for the extra precision if you're just using simple/dumb book formulas to get LAB.

See http://www.mathworks.com/matlabcentral/fileexchange/?search_submit=fileexchange&query=westland&term=westland by one of the world's leading color researchers.

[mandana]

Hi Image Analyst.' i want the most precision. how can i find the code of DeltaE 2000?

[Image Analyst]

Your comment tells me you did not look at the link to Westland's toolbox that I recommended. If you had looked at it, you would have found the function.

function [de,dl,dc,dh] = cie00de(lab1,lab2,sl,sc,sh)

By the way, I'm sure you've heard the expression "garbage in, garbage out". So, it's OK to use in some circumstances but you can't use delta E to compare colors across images, or expect to get out the same delta E like you'd get from a real spectrophotometer, unless you have calibrated your colors. Otherwise you're just getting "book formulas". If you have a color in one image taken with one camera or exposure conditions and another image from a different camera or exposure conditions, then even using the book formula is meaningless.

[mandana]

thanks so much for your help