Dear Volker. Thank you. But I'm interested in more general approach for the normalization of images. For example - the construction of normalizers for projective transformations.

If you intent to compensate geometrical distortions, the answer ist yes.

a) Usually measure some keypoints in an reference image. Methods: SURF/SIFT, Cross-correlation, etc.

b) You do a search of the keypoints (SURF, ...)  in your distorted image.

c) You compute the position differences of the keypoints in the reference and distoted images.

d) You apply a Cubic interpolation (standard algorithm) procedure on the distorted image -> output: a non-distorted image.

This approach is found in many image processing textbooks.

Dear Volker. Thank you. But I'm interested in more general approach for the normalization of images. For example - the construction of normalizers for projective transformations.

I see, you mean Projective Projections like these: http://homepages.inf.ed.ac.uk/rbf/CVonline/LOCAL_COPIES/BEARDSLEY/node3.html, right?

Yes Volker. But I want to link to the article.

Dear Professor

I know very normalization of any distribution ,mean in that we transferred any values between 0-1. but in imagine normalization is really a new thing for me. I want to learn this new idea from you.

With kind regards