Re: How to calculate two curves equidistant from a given curve?
- From: "Tony" <tonkoj@xxxxxxxxx>
- Date: 5 Sep 2005 01:11:44 -0700
Eric Gisse wrote:
> If you have a closed convex plane curve, which is also positively
> oriented [orientation doesn't matter beyond a sign change, though], the
> formula for a parallel curve that is always a fixed distance, r, away
> from the original curve, is given by this forumla:
>
> B(s) = A(s) - r*n(s)
>
> Where s is the parameter, r is the distance, and n(s) is the normal
> vector. B is called a parallel curve to A. It doesn't matter whether
> the parameterization is an arc length one or not.
>
> This was taken pretty much verbatim from DoCamro's "Differential
> Geometry of Curves and surfaces", page 47.
I see. So, in component notation it would be the formula given at
mathworld site. While this will work, the square root required for
normal vector's normalization hurts the performace of evaluating points
on such a curve.
Is there any 'reparametrization trick' that would get rid of the square
root?
Tony
.
- Follow-Ups:
- Re: How to calculate two curves equidistant from a given curve?
- From: Eric Gisse
- Re: How to calculate two curves equidistant from a given curve?
- References:
- Prev by Date: Re: what makes it true?
- Next by Date: Re: How to calculate two curves equidistant from a given curve?
- Previous by thread: Re: How to calculate two curves equidistant from a given curve?
- Next by thread: Re: How to calculate two curves equidistant from a given curve?
- Index(es):
Relevant Pages
|