1

I have an image containing:

  • a set of coordinates that act as orientation markers.
  • a set of coordinates containing data.

Let's call this image A.

This image is fed to a scanner that returns a copy of the image with certain transformations applied (rotation, scale, translation). Let's call the transformed image B. The transformation values applied are not know to me, of course.

Once I receive the transformed image (B), I can easily track the coordinates of the orientation markers and calculate the angle of rotation, scale (x,y) and translation (x,y).

Now I need to retrieve the data coordinates since I already know the transformed orientation coordinates.

If a data point was at location (10, 10) in image A, where would it be in image B? Given that all three transformations are known.

When I apply a simple matrix transformation the transformed data points I calculate are inaccurate. I tried changing the order of transformations but that seems to have absolutely no effect.

What am I doing wrong? Is it the order/sequence of transformations or something else that I'm missing?

EDIT

Please refer to this question for context.

Community
  • 1
  • 1
Raheel Khan
  • 14,205
  • 13
  • 80
  • 168
  • We would need the input and output you get. The order of the transformations definitely matters but if applied correctly I don't see why you should see differences. Does the error increase with the number of operations? IF that is the case it could be a precision problem that accumulates. – Luis Jan 22 '12 at 07:25
  • Also, if you do only one of those operations, do you see the error? – Luis Jan 22 '12 at 07:29

0 Answers0