In the DEM to TIN generation there are several steps:
1.Point Selection
2. Triangulation
Most algorithms use Delaunay triangulation, and I am OK with that for now, but as far as the point selection I am wondering if anyone has implemented anything else beside VIP and Heuristic --- I am interested in the specific algorithm(even pseudo-code) of the Skeleton Method detailed by Dr. Fowler and Little.
+56I've looked into this and don't see any other requests for implementation, or any plans to do so; which is a pity really since it looks like one of the algorithms you mention was devised at a local university.
Anyway... I doubt this is something we'd give a high priority to implementing. If you want to see it in FME officially then I suggest adding an idea on this site, and if enough users upvote the idea then we'd carry it out.
I suspect the easiest way for a user to implement this in FME is to find existing code (preferably Python) or an existing tool/application/service that can generate such a TIN. FME is well-suited to either integrating another tool (SystemCaller or HTTPCaller for a web service) or to running code if it is in Python (PythonCaller transformer). R might also be a way to integrate a solution.
I don't know what the algorithm is like for these methods, but I think it unlikely that implementing it through a series of FME transformers is going to succeed.
Thank you for the answer Mark. Indeed, the proximity of the two instances(Fowler and Little algorithm in BC Univ., an famous authoritative research that still stands today, and renown FME software) left room for me to hope you have the method already implemented in the SAFE software for DEM to TIN generation.
One of my colleague will follow up on a specific feature from FME that triggers this particular transformation, maybe that will give you a better understanding of what we are looking for.
+11Hi @7sistrunk,
I just wanted to follow up with you after our talk on Live Chat. I've spoken with our developers and I am in agreement with Mark2AtSafe; unfortunately, it doesn't seem like we are using the Skeleton Method.
Mark has offered some really good suggestions for you to potentially integrate the use of the algorithm using some of the transformers that he has listed, though we can't guarantee success in it.
Can you tell me anything more about your research project and overarching goal?
If you think that the Skeleton Method should be implemented in FME, please feel free to share your idea on the Ideas Exchange. We can't promise that it will be developed, but if there is significant interest in the idea through upvotes, it can help inform development about what users want and are interested in.
