Normalise an Uint16 image between specified limits

Hi all

I have a gray-scale image in double precision that has to be converted to uint16 and normalise the image between the limits, say, 0-4000. I could convert the image to uint16. But when I do the normalisation by dividing with the maximum intensity of the image and then multiplying it with the maximum (say 4000)that i want to have, all the pixel values are saturated. I think this operation does the rounding thing as it is an integer. could anyone let me know how the normalisation within specified limits could be done for image matrix of uint16.

Your help is much appreciated. Regards

Answer 1

[Walter Roberson]

uint16( YourImage ./ (ExistingImageMaximum .* (1+eps)) .* NewImageMaximum )

Note: with this particular formulation, 0 and the new maximum will only have 1/2 of the probability of the other bins, which has to do with uint16() rounding. If you change to multiplying by (NewImageMaximum+1) and you floor() before you uint16() then the bins should equalize.

[Kavitha Srinivasan]

Thanks for your help. Will try and get back shortly. Before that, may I know what does 'eps' stands for and how to calculate that? Regards

[Walter Roberson]

http://www.mathworks.com/help/matlab/ref/eps.html

[Kavitha Srinivasan]

Hi

It worked. Thanks