Hi! Harald JOHNSEN wrote: > The thread is mainly about land use classification, but what about roads > and rivers ? vmap0 is really inacurate and it's a pain to fly vfr > (following road). Not only a lot of features are missing but most of > those visible are off by a great distance. And then it's also difficult > to add landmarks to the scenary because there is no accurate reference > point to help positioning objects. Will we use osm in the future ? And > since osm is far from being exhaustif how to make a choice of wich one > between osm and vmap to use to generate a tile ?
Martin has investigated cooperation with OSM and IIRC the results were not satisfying, specifically regarding the willingness to cooperate, which would be necessary to feasibly integrate the data into scenery. I will have to talk to him once more what the details were. For major roads and railroads it might be possible to classify data from the imagery, but smaller roads are outside scope. Some smaller features which are hard to detect by classification but can be seen by human eye might be digitizable manually using Landsat or similar free imagery as background. It might be possible to correct parts of VMAP0. I was contacted a while ago by a researcher working on using ASTER data to determine linear features, but I haven't heard yet of their progress. The intent was to integrate OSM data as well, but also automatic detection of linear features. The concept is similar to OSM in that training data and corrections shall be supplied by webusers. Finally, VMAP level 1 is much more accurate than VMAP0 but available only for parts of the world. This is just an example of the VMAP1 _landmass_ available for the Mississippi Delta (Attention! long URL): http://mapserver.flightgear.org/cgi-bin/landcover?layer=v1_landmass&zoomdir=0&zoomsize=2&imgxy=300.0+300.0&imgext=-90.000000+28.700000+-88.740000+29.960000&root=&savequery=true&program=%2Fcgi-bin%2Flandcover&map_web_template=main.html For the Berlin scenery we have used the VMAP0-data as basis and adjusted this according to what we could see on the Landsat image. This exercise also showed, how seriously inaccurate VMAP0 is in various places: displacements by several 100m, streets which never existed, existing major roads missing or misclassified, ... Currently, there is no silver bullet regarding line features. We have different data available for different parts of the world. Actually, the scenery database is there to integrate the different data sources and we're still working on how to properly organise the processes involved. One has to connect different datasources at their borders so that the transition is as seamless as possible. On the TerraGear side I have created additions, gdalchop and ogrdecode, to enable access to all GDAL- and OGR-supported input formats. gdalchop is not yet finished (I will probably have a deeper look at that next week), but ogrdecode already has done a pretty good job for my local South Germany stuff as well as for Berlin and Oshkosh, which were built completely based on hgtchop and ogrdecode. Unfortunately, there is no OSM driver available for OGR, as a example, and the way the OSM project constantly modifies its infrastructure and data format, maintaining such a driver would be a pain in the ass. Some very dedicated guy at OSM obviously maintains an OSM-to-PostgreSQL-bridge which Martin wanted to apply to import OSM data into the scenery database. I'm not sure how usable that will be. Cheers, Ralf ------------------------------------------------------------------------- This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
