On Friday, March 3, 2000 at 4:00:00 PM UTC+8,
[email protected] wrote:
In article <[email protected]>,
[email protected] wrote:
Hi,
I'm looking for C/C++ bivariate surface fit code. I've posted a
similar
request previously but the lack of responses lead me to believe I had
not well-stated my request, or perhaps it is indeed something
difficult
to find, so I apologize if this is redundant for some, and many thanks
to those who are able to reply, esp. via email as my ISP removes
messages
every day or so.
Puzzle:
Given a set of data points randomly positioned over a 2D area and
of random value (or height, as you prefer)....
** How do I generate a regular mesh (say, 50x50 etc) of polygons
or triangles such that the surface is
A) a bicubic interpolation *through* the points with the
"tightest"
curve, and/or "least" warped by the choice of coefficients
* again, I'm looking for that regular 50x50 mesh... I did
get
a couple of emails from people prescribing tesselators,
surface
subdivision etc, but that isn't what I had requested or
what I'm
looking for since, unfortunately, it doesn't fit the
data
as prescribed :)
and/or
B) well approximated as near the points as possible with "good"
curvature, and perhaps by selection of some limit of error,
eg. possibly using least squares and I could only guess as
what kind of rational spline to use, whether bicubic or
whatever..
Is there a reason you have to form a regular rectangular mesh?
Forming the Delaunay triagulation and then interpolating that
triangulated network would be less costly in both time and memory.
I have some triangulation and linear interpolation code in C that
you might be interested in.
If you need to do a spline interpolation, again I would encourage
you to check out Dave Eberly's site. Paul Bourke has some helpful
source code too, I think:
http://www.swin.edu.au/astronomy/pbourke/
If you're interested in my code, please reply by email.
Bob
Sent via Deja.com http://www.deja.com/
Before you buy.
hi i am really interested in knowing more about your code as i am trying to do a local bicubic surface interpolation of (x,y,z) data set that would give me the resulting control points as well as the knot vecotors in the u and v direction. Hope you could
help
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)