Hi Jaya,

which software are you employing to normalize your image? If you are using Matlab, there is a really interesting database called Mathworks which contains valuable algorithms and threads. Using Matlab, I would proceed like this:

mmin = min(M(:));

mmax = max(M(:));

M = min(max(M,0),1); % [0 - 1]-normalized and vectorized values

imagesc(M); % see the normalized image

Another question is the second issue: this normalized image is not binary!!!! The matrix M is now normalized to the range [0-1] but that do not means that values of the Matrix are binary... so the output data is not similar to a binary image, to do that you need to make another transformation.

Okk,thnx

I think the problem is that you normalize (scale all pixelvalues linearly between 0 and 1) a byte, short or integer image. Converting to float or double will help.

Yes as Lambert Said converting to double will definitely do. I never had this problem.