How to Efficiently Triangulate a Concave Shape?

  • Context: Graduate 
  • Thread starter Thread starter maze
  • Start date Start date
  • Tags Tags
    Concave Shape
Click For Summary
SUMMARY

This discussion focuses on efficient methods for triangulating a closed concave shape with approximately 10,000 vertices. Key principles include minimizing sliver triangles and ensuring reasonable computation time, ideally under a day. The Delaunay triangulation method is suggested as a potential solution, particularly when combined with the convex hull of the polygon. The discussion emphasizes the importance of avoiding the crossing of existing edges during the triangulation process.

PREREQUISITES
  • Understanding of Delaunay triangulation
  • Familiarity with convex hull algorithms
  • Knowledge of computational geometry
  • Basic programming skills for implementation
NEXT STEPS
  • Research Delaunay triangulation algorithms and their implementations
  • Explore convex hull algorithms, such as Graham's scan or QuickHull
  • Learn about sliver triangle minimization techniques
  • Investigate computational geometry libraries like CGAL or Triangle
USEFUL FOR

This discussion is beneficial for computational geometers, software developers working on graphics and modeling applications, and anyone involved in mesh processing or optimization tasks.

maze
Messages
661
Reaction score
4
I am looking for a "good" way to triangulate a closed concave shape of vertices and edges.

"good" is quite a vague term, but the basic principles I'm interested in are:
1) Minimize "sliver" triangles that are very thin and contain really small angles.
2) Must be reasonable to compute. Ideally it would take less than a day to compute the triangulization for a mesh with ~10,000 vertices on a standard computer (so complete brute force is out).

Here is a diagram (initial shape in black):
http://img394.imageshack.us/img394/3077/concavetriangle.png

If the mesh was convex, a good strategy would be http://en.wikipedia.org/wiki/Delaunay_trianglulation method could work well if I had a fast way of finding the worst possible triangle among a subset of the shape, but I'm having no luck there.
 
Last edited by a moderator:
Mathematics news on Phys.org
Suggestion (it may not be very good): Form the convex hull of the polygon. Use Delaunay, being careful you don't cross any existing lines. Get rid of lines outside original polygon.
 
I think the Delaunay triangulization is unique
 

Similar threads

  • · Replies 3 ·
Replies
3
Views
4K
  • · Replies 2 ·
Replies
2
Views
3K
  • · Replies 7 ·
Replies
7
Views
4K
  • · Replies 7 ·
Replies
7
Views
4K
  • · Replies 1 ·
Replies
1
Views
4K
  • · Replies 13 ·
Replies
13
Views
10K
  • · Replies 65 ·
3
Replies
65
Views
12K
  • · Replies 1 ·
Replies
1
Views
3K
  • · Replies 1 ·
Replies
1
Views
4K