Question

I want to accumulate the edges of triangular model/mesh. For example a triangular cube has 18 edges.

It looks so simple and easy but literally its too complicated.
I have all adjacent information of a triangle. For example I have adjacent vertices and triangles, and I know which triangles have common edge but the problem is how to extract that common edge between two triangles.

One thing that needs to consider is there should not be any duplicate index of edge’s point/vertex.

First thing is how to count the total number of edges. Remember a Cube has 18 edges.

I tried a lot but gave up now. 🙂
any idea?
.

Update 1:

ok I have a triangle T[i] (index v1, index v2, index v3) which has three edges v1v2, v2v3, v3v1.

and I have a neighbor triangles of all edges.

T[i].index_of_sharedTri1_with_edge_v1v2, 
T[i].index_of_sharedTri2_with_edge_v2v3, 
T[i].index_of_sharedTri3_with_edge_v3v1; 

now which loop I need to make that can extract unique edge? What information I need to compare? Do I need to compare edge vertices, shared tri index, or what ? I tried with many ways but it so complicated.

Update 2:

GLTris *e = new GLTris[nb_Tris*3];
int n = getTotalEdges_Sorted(indices, nb_Tris, e);
cout<<n<<endl;
int ne = RemoveDublicatesFromAnSortedEdgeArray(e, n);
for(int i=0; i<ne; i++)
   e[i].Cout();

I have tried this and it works perfect but I need to figure it out whether it will work on all types of meshes and is it a efficient way?

Answer 1

If you have the number of triangles, and the adjacency information for each, then the number of edges is simply:

3 * num_triangles – num_shared_edges

Where num_shared_edges can be accumulated by iterating over each triangle and adding up the number of adjacent triangles. Then divide by two, as you’ll have counted them all twice (this obviously assumes you have no more than two triangles per edge)

If you want to build a list of unique edges, you’ll need to keep track of them, preferably in some easily indexable structure like a map.

The procedure is roughly as follows:

1. Define an edge as being two indices (i1, i2). Order those indices such that i1 < i2. Now shared edges are guaranteed to match.
2. Make some kind of list structure that can hold these edges. You need to be able to quickly find an existing edge by its indices, so you want a structure that is searchable, such as some kind of tree. If you were using c++, I’d suggest using a map with the i1,i2 pair as a key.
3. Iterate over all triangles, adding 3 edges for each. Do not add the edge if it already exists.

That’s all there is to it.