Hi, Interesting task. It principle it does not sound very difficult but still I do not know how to make it exactly. Rectifying aerial images is rather similar task actually. First you have just an image what containts a hundred million pixels. Then you give ground control points for a handful of image pixels, and warping processes not only those but also all the pixels in between GCP's.
>From OpenJUMP tools I guess warping tool might be useful. Affine >transformation does not necessarily suit this case, because each polygon can >be distorted in an individual way. You could also have a try and ask your question from gdal-developers mailing list. Try to describe your problem and make it understandable that you would need to do something that is close to warping raster images which have been georeferenced with ground control points, but with vectors. If you would like to get it accurate to few inches, you'll need ground control points accurate to better that that. It means survey grade measurements. If ten inches is enough then a good aerial image could be used as reference. If you are very lucky you may find an existing image from mapping agency archive for a reasonable price. I guess it does not harm if the image is a few years old. -Jukka- Kurt Heston wrote: No aerial photo, drawn freehand. If you're reading this, you're probably thinking, "there had to have been something they used to reference this" but trust me, there wasn't. One look at the data and you would know for sure there wasn't. The maps were drawn to look as though 3 sections of the park that are actually separated by city streets are right across the street from one another. The trouble is, one of these sections is actually 3 blocks away. If there was any hope of finding the way this data might have been georeferenced originally, believe me, I'd have found it. No one is more upset than I am about the prospect of having to re-create this data. I'm just the guy charged with making the last guy look as though he did a great job. Thanks for the input. I appreciate the suggestions. Rahkonen Jukka wrote: > Hi, > > You told the polygons in the shapefile have been hand drawn to look good on > screen. Thats's how people use to digitize things on top of georeferenced > images. How your data were actually captured? Did you have some scanned map > or aerial photo as a background image? > > -Jukka Rahkonen- > > > Kurt Heston wrote: > > Stefan, > > I think you do understand the problem as I've described it and I think I > am following your geo-speak. > > The polygons are primarily 3x9 foot spaces in a ~100 acre cemetery. So, > accuracy within a few inches would be ideal (though I agree this will be > a TALL order and likely require the rest of my life gathering sample > points). If that's impossible, something closer than what I have right > now would be nice. > > I'll look into warping and also affine-transforms and see if I can speak > more intelligently about what my specific problems are and how to go > about fixing them. > > Thanks for the help! > > > > Stefan Steiniger wrote: >> Hei Kurt, >> >> yeah.. I am also not quite sure how to help you. >> However, if you have reference points that are in your target coordinate >> system an in the dataset that you would like to transform, then you >> could try to use the Affine-Transformation Tool (>Tools>Transform). I >> think a HowTo can be found in one of the original JUMP documentations. >> >> However, I am not realy sure if that brings you the accuracy you need, >> i.e. you didn't wrote >> - what is your current Coordinate/Cartographic Projection System (is it >> in meters or in lat/lon = degrees?, is it GPS data) >> - what is your target Coordinate/Cartographic Projection System (short >> sometimes: CRS for Coordinate Reference System), >> - what transformation accuracy is needed (m, submeter, 100m?) >> - ... ?I forgot that? ... >> >> to develop/use a transformation procedure would depend on all those things. >> Sorry, I may have chosen a bit a too technical language - but this is >> somehow not avoidable with respect to your question (and if I understood >> it correctly) >> >> stefan >> >> Brent Wood schrieb: >>> Hi Kurt, >>> >>> If you can work out what the projection system is for your data, then >>> Proj.4 can do the conversion for you. >>> >>> Note that it is likely to be easier to use ogr2ogr (part of GDAL) as >>> this can use the proj.4 libraries to reproject data, but can also >>> read/write shapefiles so you could generate reprojected shapefiles >>> directly from your original shapefile. >>> >>> The key to thi working is knowing the coordinate reference system of >>> your source data. >>> >>> If this infomation is not available, then some sort of manual >>> coordinate transformation may be possible, based on your suggested >>> approach of working from visually determined points, but I'd use this >>> approach as a last resort, especially if any sort of accuracy is >>> required. >>> >>> see: http://www.gdal.org/ >>> >>> >>> >>> Cheers, >>> >>> Brent Wood >>> Brent Wood >>> DBA/GIS consultant >>> NIWA, Wellington >>> New Zealand >>>>>> Kurt Heston <[email protected]> 05/11/09 10:22 AM >>> >>> I've inherited a set of shape files that are not georeferenced. The >>> upper left corner of the extent is on 0/0 lat/lon. So, when I try to >>> view them after translating to KML, for example, they show up in the >>> middle of the Atlantic Ocean. >>> >>> Is there any way in OpenJump that I can fix them? There's about a 130k >>> polygons involved with lots of attributes that I'd like to forego >>> re-creating if I can. >>> >>> If I use a survey-grade GPS or Google Earth to match a bunch of points >>> in these shapefiles to their true locations on Earth, can something >>> move/stretch/skew/scale them to their earthly positions? This HAS to >>> have been done at some point with all the non-geo CAD stuff that existed >>> b4 geo-spatial tooling took hold, right? >>> >>> I'm hoping there's a tool I've missed (in much Googling) that I can use >>> to correlate the two. I'm guessing that if such a tool is available, >>> it's going to take A LOT of point matches to fix the files, but that >>> scares me less than re-creating the data in its entirety. >>> >>> Can anyone send me in the right direction here? >>> _______________________________________________ >>> jump-users mailing list >>> [email protected] >>> http://lists.refractions.net/mailman/listinfo/jump-users >>> >>> NIWA is the trading name of the National Institute of Water & >>> Atmospheric Research Ltd. >>> _______________________________________________ >>> jump-users mailing list >>> [email protected] >>> http://lists.refractions.net/mailman/listinfo/jump-users >>> >>> >> _______________________________________________ >> jump-users mailing list >> [email protected] >> http://lists.refractions.net/mailman/listinfo/jump-users > > _______________________________________________ > jump-users mailing list > [email protected] > http://lists.refractions.net/mailman/listinfo/jump-users _______________________________________________ jump-users mailing list [email protected] http://lists.refractions.net/mailman/listinfo/jump-users _______________________________________________ jump-users mailing list [email protected] http://lists.refractions.net/mailman/listinfo/jump-users
