Re: Delaunay: Determinant test
- From: "Pixel.to.life" <pixel.to.life@xxxxxxxxx>
- Date: 13 May 2007 15:05:21 -0700
On May 13, 1:07 pm, all...@xxxxxxxxxxxx wrote:
Hi all.
I am struggling a bit to understand how the determinant test for
the Delaunay triangulation works.
Assume the triangle T(A,B,C) exists, and that a point P is introduced
to be included in the Delaunay triangulation of the set {A,B,C,P}.>From various sources (web, Preparata & Shamos, de Berg & al), I
have seens statements to the effect that "If the determinant D = ...
is positive, the Delaunay triangulation is <this>, else it is <that>."
I can get this up and running as far as one triangle goes: Given
the three points A, B and C, one knows that there can be constructed
a circle with radius R and centre S such that all points are located
on the circle.
The above is easy to formulate as a matrix:
(x_A-x_S)^2 + (y_A-y_S)^2= R^2
(x_B-x_S)^2 + (y_B-y_S)^2= R^2
(x_C-x_S)^2 + (y_C-y_S)^2= R^2
A little manipulation and one finds a matrix expression which
can be solved for x_S, y_S and R. From there on, I am stuck:
1) How does this idea extend to testing two triangles?
2) How does the test boil down to testing the sign of the
determinant?
I am probably just blind, since I haven't used determinants since
the intro course on linear algebra I took 15 years ago.
Rune
Hi, Rune.
Your honesty is appreciable. There is nothing wrong in admitting a
long lost concept, or lack of knowledge. Thats how we learn and share.
Have you come across the "Algorithms" section of this article?
http://en.wikipedia.org/wiki/Delaunay_triangulation
You see a matrix representation used to test in a 2D case whether to
add a point D as a new point for triangulation.
Given three points A, B, C forming a triangle with edges ordered
counter-clockwise, you introduce a fourth point D.
When you compute the determinant of that matrix, it is positive if and
only if D lies within the circumcircle of triangle ABC.
Now based on whether or not D lies inside the circumcircle, you take
the next step appropriately ( would you want to introduce a new point
inside ABC? would it be optimal? )
Hope this helps a bit,
Pixel.To.Life.
[ http://groups.google.com/group/medicalimagingscience ]
.
- Follow-Ups:
- Re: Delaunay: Determinant test
- From: allnor
- Re: Delaunay: Determinant test
- References:
- Delaunay: Determinant test
- From: allnor
- Delaunay: Determinant test
- Prev by Date: Delaunay: Determinant test
- Next by Date: Re: Delaunay: Determinant test
- Previous by thread: Delaunay: Determinant test
- Next by thread: Re: Delaunay: Determinant test
- Index(es):
Relevant Pages
|