[Openstreetmap] Finding roads in a cluster of GPS points (Was: How do we handle large amount of GPS points?)

Matt Amos matt at matt-amos.uklinux.net
Mon Nov 22 23:39:41 GMT 2004


On Monday 22 November 2004 21:37, Petter Reinholdtsen wrote:
> [Matt Amos]
> > i was thinking the same thing: plot HDOP circles around points
> > and colour the lines according to "goodness of fix". that could
> > be a very arbitrary function of number of satellites, fix, etc...
>
> What about creating a matrix for the area in question, and for each
> GPS point increase the value by one of all matrix points within a
> given radius of the coordinate.  The radius could be a function of
> the accuracy of the point.  This should give a matrix with peaks
> and "hills" where the roads are.  One could then trace the hills,
> and find roads that way.

thats certainly possible. it would be a good idea when tracing very 
heavily travelled roads, as its a nice intuitive system.

what i was trying to explain was the method of presenting gps point 
accuracy to a human user, so there would be an intuitive way of 
recognising the difference between an innaccurate and accurate route. 
then the user can make an informed choice about the placement of the 
road  (overriding the automatic placement if necessary).

> Not sure how effective this would be.  One could try it on limited
> areas first.

i'm wary of using any sort of raster method to help trace our vector 
data, after all; the gps points are vector-based and the final routes 
are vector-based. placing a raster method in between just reduces the 
accuracy of the final vector data to the raster pixel size.

cya,

matt
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 190 bytes
Desc: not available
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20041122/6676724a/attachment.pgp>


More information about the talk mailing list