Jump to content

  • Log In with Google      Sign In   
  • Create Account


Curve fitting with minimum oscilation


Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

  • You cannot reply to this topic
2 replies to this topic

#1 goruka   Members   -  Reputation: 125

Like
0Likes
Like

Posted 13 June 2012 - 10:54 PM

Hi! I've been researching curve interpolation algorithms for a while, and I have the following problem.
All the classic ones like cubic spline, hermite, catmull rom, etc add small oscilations near the control points, and I don't want this. Are there other curve-fitting algoritms that, no matter if they need more control points than 4, they don't produce such oscillations?

Sponsor:

#2 jjd   GDNet+   -  Reputation: 2075

Like
0Likes
Like

Posted 14 June 2012 - 07:15 AM

Hi! I've been researching curve interpolation algorithms for a while, and I have the following problem.
All the classic ones like cubic spline, hermite, catmull rom, etc add small oscilations near the control points, and I don't want this. Are there other curve-fitting algoritms that, no matter if they need more control points than 4, they don't produce such oscillations?


I'm not sure what you mean by oscillations in the curves -- perhaps you could provide a picture? In that regard, the Catmull-Rom spline should be pretty good and minimizing unnecessary curvature in the curve. Are you sure that it is not a problem with your implementation?

-Josh

--www.physicaluncertainty.com
--linkedin
--irc.freenode.net#gdnet


#3 Álvaro   Crossbones+   -  Reputation: 12912

Like
0Likes
Like

Posted 14 June 2012 - 07:23 AM

So-called "natural" cubic splines (where the second derivative of contiguous segments matches at the common node and the second derivative at the ends is 0) have the least "curvature" among all interpolating curves, in some sense that can be made very precise (minimum integral of the square of the second derivative). So I would give them a try. If you are not happy with the results, you must have some other criterion in mind, so it would be good to see some examples.




Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.



PARTNERS