Hello all,

 

I hope you can help me, I’m trying to group 3D point data in clusters according to the distance between points. I.e. different groups (or components) are separated by a minimum threshold distance.

To do this I am creating a boost graph (using the Boost Graph library), adding vertices (with 3D point information) and adding edges between nodes of the graph that are within my threshold distance.

 

However, when I find the connected components on the resulting graph I’m getting an incorrect answer. The vertices of the graph are not grouped correctly. I am comparing my results with a Matlab script (and their proprietary programs) which correctly separates the points.

 

I have a stackoverflow post with more detail (and no answers) http://stackoverflow.com/questions/27001402/connected-components-boost-c.

 

If you are able to help in any way I’d be very appreciative. I really don’t want to use compiled Matlab for this solution.

 

Thanks

 

Simon Choppin