HI JK

That makes your problem clearer, one thought, the geometry looks the same in both pictures so what did you do to

make one correct and the other not ?.

I went here

https://www.sciencedirect.com/science/article/pii/S0377042798002234and had a look at this paper.

It seems to address cases where you have more than two choices and uses a simplified method.

as I understand they seem to be addressing this problem.

Here is a Quote from the paper-->

270 R. Rarnamurthy, R.T. FaroukilJournal of Computational and Applied Mathematics 102 (1999) 253-277

A Newton-Raphson iteration scheme, analogous to that described above, can be developed from

these equations for refinement of type (3~) bifurcations.

Finally, consider the case of a bifurcation b* with more than three distinct footpoints on the

domain boundary S. Such bifurcations are “exceptional”, but do not require any special treatment -

the methods discussed above suffice for computing their exact coordinates. To see this, suppose that

b* has distinct footpoints on four segments sl, s2, s3, s4 of S. Now when three of these boundary

segments, sl, ~2, s3 say, have been included in the boundary segment set, 6* will be identified as a

“generic” bifurcation point of type (3). At this stage, depending on the type of b*, the appropriate

algorithm discussed in this section may be employed to locate the exact coordinates of b*. The

subsequent inclusion of s4 does not alter the coordinates of 6* in any way. Thus, due to our strategy

of including one boundary segment of S at a time, and because of the fact that each boundary

segment is “simple”, we are guaranteed to capture all exceptional bifurcation points having more

than three disticnt footpoints on the domain boundary.

6. Medial axes

We now consider the construction of the medial axis of a planar domain D from the Voronoi

diagram of the domain boundary S. The medial axis may be formally defined as follows [6, 321:

Definition 6.1. The medial axis of a planar domain D with boundary S is the closure of the set of

points in D that have (at least) two distinct footpoints on S - i.e., it is the closure I9 of the point

set

I think you are right track trying to find a better library, and I will look into the problem a little deeper in the next few days to see if i can find a library for you to use (this may take a little time as I'm in the middle of my turret project)

making the overload mechanisms and although it looks simple they rely on very close tolerances and I've already Trashed a couple of them.

Dave