1

I have a series of 2D images which i have fitted a 6th degree polynomial into:Polynomial fit

I want to transform every 2D image into a line with respect to the polynomial and stack the lines so i get a 2D image (basically the job is to transform 3D CBCT-scan to 2D OPG image). I have tried just picking the pixels on the polynomial but that lead to a stretching image like this: enter image description here

how can i transform image pixels on the polynomial right?

Pariya Mehrbod
  • 31
  • 6
  • How are you projecting the lines into the plane? – Mauricio Cele Lopez Belon Aug 30 '20 at 22:32
  • For each point with the coordinates (x,y) on the fitted polynomial i get the value of the pixel with (x,y) coordinates so for each plane i have a line. – Pariya Mehrbod Aug 31 '20 at 07:12
  • So the lines are not projected to the plane, they are mapped to the plane. And the mapping function is distorting the line (stretching or compressing it). So, going back to your task you have a stack of lines, each line is 1D function but you can embed it in 3D space, so by stacking you lines you actually have a 2D surface embedded in 3D space. Did you try doing a perspective projection of that surface to the plane? – Mauricio Cele Lopez Belon Aug 31 '20 at 09:21
  • exactly, i have not tried perspective projection but i don't know how that should be applied here, could you please guide me? – Pariya Mehrbod Aug 31 '20 at 11:46
  • You give too little information about the kind of "projection" you want. Can you show at least the expected result ? –  Aug 31 '20 at 14:24
  • the expected result is a full view of all teeth like an opg image (not stretching or compressing) – Pariya Mehrbod Sep 01 '20 at 15:11
  • you should share sample input image/data and ideally also wanted output (does not need to match the input) so we can better asses the needed transformations ... because just the wrongly stretched result is not enough for non common input cases like this... – Spektre Oct 01 '20 at 07:50

0 Answers0