Hi everyone :) i take some times to look more in the data of urbis, i also played with qgis to try to extract some informations... i started with buildings and addresses as i think it's the weakest point in brussels...
So here is my take on etterbeek (as it's a small one): I split it in 3 files : - buildings without addresses - building with 1 address (address on the building shape) - building with more than 1 address (address as diffrent points) What do you guys think? before uploading there is still some work to do : - check for duplicated node or self intersecting buildings - check for existing buildings / address in osm Any comments? http://my.bmaron.net/public.php?service=files&t=c73bc235b68982c874d6e29f9223a741 On Fri, May 3, 2013 at 10:31 AM, eMerzh <merz...@gmail.com> wrote: > Hi , > Thanks for you researches Jo... i'm currently reading the user guide given > upper, and it may helps you to understand what's what :p > > yesterday Cquest from osm-fr gave me his extract of addresses so it must > be possible some way ... (see here > http://my.bmaron.net/public.php?service=files&t=7264154f5a483352ebe0846fdeeeabbb) > > For the integration ... as it seems that the datas are far better than > what's already in osm ( a lot of buildings where drawn by hand on top of > bing ....) > and addresses are far from complete ... > my opinion is that we may try to first "integrate" buildings then address > (address seems to be easier i think it's better if we could link them to > the building). > Another place where an import will be hard but we can still do smth is > roads names. > > Then of course there is a lot of other things we can start thinking ... > like pharmacy , ... > > Btw, > how do we manage the attribution ? a text in the wiki pointing to URBIS? a > source=* on every object ? or a source=* on the changeset? > > > On Fri, May 3, 2013 at 10:27 AM, eMerzh <merz...@gmail.com> wrote: > >> Hi , >> Thanks for you researches Jo... i'm currently reading the user guide >> given upper, and it may helps you to understand what's what :p >> >> yesterday Cquest from osm-fr gave me his extract of addresses so it must >> be possible some way ... (see in attach) >> >> For the integration ... as it seems that the datas are far better than >> what's already in osm ( a lot of buildings where drawn by hand on top of >> bing ....) >> and addresses are far from complete ... >> my opinion is that we may try to first "integrate" buildings then address >> (address seems to be easier i think it's better if we could link them to >> the building). >> Another place where an import will be hard but we can still do smth is >> roads names. >> >> Then of course there is a lot of other things we can start thinking ... >> like pharmacy , ... >> >> Btw, >> how do we manage the attribution ? a text in the wiki pointing to URBIS? >> a source=* on every object ? or a source=* on the changeset? >> >> >> >> On Fri, May 3, 2013 at 10:07 AM, Jo <winfi...@gmail.com> wrote: >> >>> Hi eMerzh, >>> >>> The data is provided in different ways. I started by looking at the >>> topology files, but everything is totally fragmented in there and I can't >>> seem to find the glue to tie it all together once again. One would have to >>> 'reconstruct' the front, sides and backside of the houses and combine with >>> house numbers. >>> >>> The adm files seem to be more accessible. PostGIS has a plugin which >>> enables to import dbf and shp files. So from the Urbis site we would >>> download the data in SHP format. >>> >>> Unzip and then point Postgis to the DBF files. Don't forget to change >>> the character encoding to LATIN1 (or set it to that for the whole DB, but >>> then it needs to be converted to UTF-8 later on). >>> >>> Once it's all imported, start QGIS and connect to the DB. Now it's >>> possible to visualise the data. >>> >>> Urbadm-bu contains the buildings. >>> >>> What we should decide is what do we want to use for import/integration? >>> >>> Many of the buildings in Brussels are already drawn. Are we going to >>> replace them with this data? Does replacing mean: throw away the nodes and >>> start over, or do we try to keep the nodes and "simply" change their >>> positions? (Not so simple to code, but probably not impossible). >>> >>> What is probably simple is to create an OSM-file with all the >>> housenumbers. Then it's still a lot of manual labour to put them in and >>> verify with what we already had. >>> >>> Creating OSM files with all the building outlines would be a bit harder >>> to accomplish. I've been trying to decipher how the data is organised in >>> the tables for the past few hours... >>> >>> I can't seem to find the link between the buildings and the >>> housenumbers/streets, but maybe I should have a fresh look at it, once the >>> headache goes away... I do think it's in there somewhere. >>> >>> Jo >>> >>> >>> >>> _______________________________________________ >>> Talk-be mailing list >>> Talk-be@openstreetmap.org >>> http://lists.openstreetmap.org/listinfo/talk-be >>> >>> >> >
_______________________________________________ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
