3

I am calculating points on triangulated grid(blue) and since I had idle cores I added points (red): enter image description here

To do this I triangulated the original grid and then added points in the centroids of the biggest triangles. As you can see this does not result in a nice grid, but with some very uneven triangles.

How do I determine a "nice" point to add a new point, without shifting any of the old? I feel that I can do this much better manually. Are there any known schemes?

Stein
  • 263
  • Can you apply Laplacian smoothing to the red points, moving each red point to the centroid of its star? – lhf Jul 12 '17 at 14:43
  • When I learned about the FEM we touched on adaptive schemes that refine triangles above a certain area. You can maybe search some literature for that. – Sean Roberson Jul 12 '17 at 14:44
  • @lhf I am using the centeroid of the triangle. What do you mean by star? – Stein Jul 12 '17 at 15:49
  • @SeanRoberson I am using the Triangle library. They offer a feature like this, but to my understanding this means moving the blue points. Which is fine in FEM, but not in my case. – Stein Jul 12 '17 at 15:50
  • The star of a vertex is the set of vertices that are adjacent to it. Your picture has red vertices that have more than 3 adjacent vertices. – lhf Jul 12 '17 at 17:16
  • The usual process for getting, out of scratch, the less elongated triangles, is to produce a Delaunay triangulation. But your problem is rather different. – Jean Marie Jul 12 '17 at 18:45
  • @JeanMarie but you can flip bad edges without moving the blue points. – lhf Jul 12 '17 at 18:51
  • @JeanMarie I am using Delaunay triangulation to create the connections, the question is where to add new points without moving old ones. – Stein Jul 12 '17 at 19:06

0 Answers0