| 2D interpolation with sparse data [message #54101] |
Tue, 22 May 2007 07:38 |
Ken G
Messages: 14
Registered: September 2006
|
Junior Member |
|
|
Here's an interesting interpolation problem.
Suppose I have a coarsely sampled 2D dataset--an image. There are
several ways to fill-in the missing data, including TRIGRID and
TRI_SURF, etc. What I find though, is that these methods can introduce
severe artifacts due to the nature of the triangulation.
This example figure here shows the problem clearly:
http://goldberg.lbl.gov/newsgroup/interpolation_problem.jpg [28k]
My original image has simple, horizontal bands with no vertical
features. My sparse sampling is collected at striped angles, as you
can see. I realize that these interpolations aren't 'wrong' per se:
the way in which they are triangulated strongly affects the final
result.
Short of re-writing my own triangulation routine, I am wondering if
there is already a way that I can tell TRIANGULATE to prefer
triangulation along the x-direction, for example, which in this case
would solve the problem. Or if there is another built-in routine that
might work better for me?
I have tried using various Fourier filtering ideas that didn't work
out as well as I had hoped. I also tried rotating my data-set in
various ways, triangulating, and then rotating back. So far, those
ideas haven't worked either.
Any ideas?
Thanks,
Ken G
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
