Normalization of the image - this is the problem of determining the parameters of geometric transformation, which has undergone an image in order to compensate them.
Are there studies on the subject?
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?
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
- Badges
- Science topic
- Similar topics
- Computer Science
- Images