Natural neighbor interpolation - help with C translation

Ideas, enhancements, feature requests and development related discussion.

Natural neighbor interpolation - help with C translation

Postby sascha » Tue May 27, 2008 8:01 pm

I'd need help with something called natural neighbor interpolation. I need it for multidimensional morphs because the current inverse distance weighting produces quite poor results. See this thread for what I'd like to do. Some papers and articles about natural neighbor interpolation for 2D data can be found on the Internet, but here's a paper by David F. Watson describing an algorithm that works in n dimensions, together with the C source of the implementation.

If someone could help me translate the C code into Java, I'd appreciate it!
PDF
C source

Thanks a lot!
sascha
Site Admin
 
Posts: 2792
Joined: Thu May 20, 2004 9:16 am
Location: Austria

Re: Natural neighbor interpolation - help with C translation

Postby Paolo » Wed May 28, 2008 6:01 pm

I will offer my help but not before 15th june, if it's still ok to you. I'll wait that time before starting (in case someone has more spare time than me).
Paolo
 
Posts: 19
Joined: Tue Feb 20, 2007 10:12 pm
Location: Italy

Re: Natural neighbor interpolation - help with C translation

Postby sascha » Thu May 29, 2008 7:33 am

Thanks for your offer, Paulo!

I'm currently implementing a slightly different approach: It too uses Delaunay triangulation (or its equivalent in higher dimensions). To find the interpolation weights for any position, it searches the simplex that contains the point in question and then performs linear interpolation. This should work well with any number of dimensions, but won't produce smooth transitions due to the linear interpolation.

I'll have to check whether my approach works and if natural neighbor interpolation produces better results. I'll use the command-line C version of the algorithm to run some tests - should the results be disappointing there's no need to convert the algorithm into Java.

There's no hurry too, I can keep coding using the inverse distance interpolation that's already in place. So I'll get back to you on June 15th.
sascha
Site Admin
 
Posts: 2792
Joined: Thu May 20, 2004 9:16 am
Location: Austria

Re: Natural neighbor interpolation - help with C translation

Postby sascha » Thu Jul 17, 2008 10:40 am

I've tested various algorithms and came to the conclusion that polyharmonic splines are the best approach, so I'm going to implement morph interpolators that way. So currently there's no need to continue with natural-neighbor interpolation - it might change in case my approach doesn't work out, although I don't think so. Anyway, thanks for the offer.
sascha
Site Admin
 
Posts: 2792
Joined: Thu May 20, 2004 9:16 am
Location: Austria


Return to Development

Who is online

Users browsing this forum: No registered users and 1 guest

cron