
Good idea! It is also possible that the --process-exits and --process-destination options might create several ways with short arcs. I should change it so that both options do not create ways shorter than the shortest allowed arc. WanMil
Hi Andrzej,
thanks for the hints. I just wonder if anybody ever tried to use a very simple trick: If I remember correctly, we calculate and write the arc length to the img file. What happens if we just make sure to write the minimum length (e.g. 3.5m), no matter how short the arc really is?
Gerd
popej wrote
Hi,
I'm doing similar processing for cgpsmapper. My algorithm is following:
First I try to remove very short lines. I don't remove lines which are a part of roundabout, restriction or road sign. I check ending nodes of short lines and don't remove lines if any node is external (on tile boundary).
I merge nodes and move them to the middle of removed line. But if a node connects to multiple short lines, I leave it at original position (this is mostly to preserve integrity of network).
Then I try to extend remaining short lines. I move both ending nodes unless one of them connects to multiple short lines.
I'm working with floating points coordinates but I round coordinates to 24-bit equivalent to calculate length. I remove only the shortest lines, where difference between ending coordinates is no longer than 1-bit in 24-bit range. I simply hope there won't be problems like removing whole bridge.
I don't know what is real Garmin condition for short line. Cgpsmapper uses distance 2.4m. I would rather expect that condition is based on 24-bit coordinate grid. Would be good to know precisely, when line is too short, this would minimize the number of changes.
-- Best regards, Andrzej _______________________________________________ mkgmap-dev mailing list
mkgmap-dev@.org
-- View this message in context: http://gis.19327.n5.nabble.com/special-cases-with-short-arcs-tp5778708p57788... Sent from the Mkgmap Development mailing list archive at Nabble.com. _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://lists.mkgmap.org.uk/mailman/listinfo/mkgmap-dev