Hi, I have a large number of SpatialPixelsDataFrame (or a grid with points that can have either value 0 or 1). The pixels/points all cluster together, so if there is a pixel at location (x, y), there is guaranteed to be a pixel at an adjoining location also. So each SpatialPixelsDataFrame contains one and only one contiguous cluster of pixels (points with value 1) and the rest of the pixels have value 0. There may be holes in the clusters.
What I thought would be simple, but can't seem to program my way out of, is to get a polygon demarking the outline of the cluster. I know I could use chull, but it will draw a circle around a horseshoe-shaped cluster (quite reasonably, as it is intended to make convex hulls). I would prefer a polygon shaped like the outlines of a horseshoe, and I figure this should be possible given that the data are gridded, so no need for a convexity condition. Also, I'd like to trace the outlines only, not any holes in the clusters. This must be a reasonably well known problem to geographers and others, but as I said, I can't figure it out. Can anyone help with ideas, pointers, code or anything? Thanks very much in advance! Bjarke. _______________________________________________ R-sig-Geo mailing list [email protected] https://stat.ethz.ch/mailman/listinfo/r-sig-geo
