An EMST is a minimum spanning tree where the graph is the complete set of Euclidean distances between the points.
It is well-known that it is a subset of the Delaunay triangulation of the point set and can be constructed for it in linear time.
The Delaunay triangulation can be optimally computed in time O(N Log N) using the sweepline paradigm (Fortune) or Divide & Conquer (Lee and Schachter).
I know that a direct implementation of the EMST is possible without explicit construction of the triangulation, at least by the sweepline approach, but I couldn't find a suitable reference.
Can any of you point me to a publication of a straight but efficient EMST algorithm ?
Please note that I am not interested in the rectilinear EMST problem.