[OSM-dev] Doing street names from aerial imagery

John Robert Peterson jrp.crs at gmail.com
Wed Mar 24 12:08:52 GMT 2010


So we can assume that a bounding area for what we are interested in will be
supplied;

Assume that there are no oneway systems in place unless that data is already
on the server;

Assume that all roads are equally passable unless otherwise staed in the
database (either by speed restictions or other metadata);

Could road passability be calulated usefully from road cureyness? or would
it be bust simply to use length?

to summerise what I have read so far:

There are a set of nodes (each intersection) a set of ways (each of which is
linked ot 1 or more nodes) and a separate routing table between nodes (to
contain routing metrics)

The routing table would be a sparse data stucture (liked list or database
reference system etc) and would be asemetric to accomodate one way systems
etc.

Each way only need one of it's nodes to be visited.

We are working out multiple predetermined psudo-optemistic routes that meet
the above.

I've studied this stuff (at undergrad level) but you'd need ot be a smarter
person than me to figure that one out... (not to put a dapener on things,
there are many people who are smarter than me...)

JR


On 24 March 2010 11:50, John Smith <deltafoxtrot256 at gmail.com> wrote:

> On 24 March 2010 21:47, andrzej zaborowski <balrogg at gmail.com> wrote:
> > geometry.  A complete algorithm would have to go through each possible
> > combination of all the routing parameters (oneways, traffic calming,
> > label on one end, label on the other end, etc etc) and find a solution
>
> Depending on the resolution of the imagery, you can some times work
> out direction and one way due to arrows painted on the road surface.
> Same with traffic calming...
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/dev/attachments/20100324/5449c31d/attachment.html>


More information about the dev mailing list