3

OK. Let's say that I have 2 images. The first is a set of lines that form a grid with a circle. The second is the same grid and circle DISTORTED. The second image is the distorted result of the first image when it is stretched over a 3D solid.

What I want to do is morph the circle in the second distorted image to where it would be in the first image before distortion. Stated differently, I want to predict what shape the circle needs to be before distortion so that when it is distorted it ends up looking like the perfect circle in the first image.

The grid is only there as an aid, I am really interested in the circle. The circle is just an idealized image. In real life the "circle" represents a graphic of some kind. The two images would be actual scans from the sheet before and after distortion. The distortion is also just idealized. In real life it is more complicated than this.

I am open to commercial software or using Matlab or other software to write something. I just have no idea where to start on this one. All ideas are appreciated.

BTW...here is a post that is somewhat similar in nature.

Original - Undistorted image: Original - Undistorted

After distortion: Distorted

Community
  • 1
  • 1
Eric Snyder
  • 1,816
  • 3
  • 22
  • 46
  • 1
    this is not exactly a programming question... – A. Donda Feb 13 '14 at 19:01
  • 1
    Is there a better place for this kind of question? – Eric Snyder Feb 13 '14 at 21:15
  • 1
    As far as I know there is no better place to ask this kind of question. – Alceu Costa Feb 13 '14 at 23:15
  • Just to confirm, you are given two shapes: the regular shape X and the distorted shape X', where XT = X' and T is a transformation. Now your goal is to find a third shape X'' such that X''T = X. Is that correct? – Alceu Costa Feb 13 '14 at 23:24
  • Yes. Perfect. I think you understand. – Eric Snyder Feb 13 '14 at 23:27
  • @Eric, maybe Signal Processing SE (includes image processing), maybe Math SE. But I'll admit the match is not much better. – A. Donda Feb 15 '14 at 15:24
  • Eric, I'm seeing several parts to this. Determining what the warpage is, and undoing its effect on the image. If you had the x and y positions of the crossings in the deformed image, could you transform them back to the original image (or an approximation thereof)? What is a good-enough approximation? Can I treat each small square as having a linear transform and ignore the curve in the edges? – EngrStudent Mar 09 '14 at 00:14

0 Answers0