Digitizing and Geocoding Old Maps? 235
Posted
by
timothy
from the walking-directions-rivendell-to-el-dorado dept.
from the walking-directions-rivendell-to-el-dorado dept.
alobar72 writes "I have quite a few old maps (several hundreds; 100+ years old, some are already damaged – so time is not on my side). What I want to do is to digitize them and to apply geo-coordinates to them so I can use them as overlays for openstreetmap data or such. Obviously I cannot put those maps onto my €80 scanner and go. Some of them are really large (1.5m x 1.5m roughly, I believe) and they need to be treated with great care because the paper is partly damaged. So firstly I need a method or service provider that can do the digitizing without damaging them. Secondly I need a hint what the best method is to apply geo coordinates to those maps then. The maps are old and landscape and places have changed, it maybe difficult to identify exact spots. So: are there any experiences or tips I could use?"
Contact your local universities (Score:3, Interesting)
Really, history majors will love this stuff. Giving them maps and a concept of Google maps overlays for real time comparisons to modern maps will likely be a capstone project for some undergrad.
A few years ago while working for the State of Wisconsin's Board of Commissioners of Public Lands we worked with the University of Wisconsin: Madison to get all of the original land plat maps of the state digitized, indexed and search-able. Same type of deal, huge maps on really old paper that had to be vault kept with humidity and temperature controls.
-Rick
Digicam? (Score:3, Interesting)
Focus on the preservation of the imagery first, obviously, because once that's gone it's gone forever.
The cheapest option is a large-megapixel digicam known for good image quality. SLR would probably be a good bet. You can take multiple images and stitch them together without too much trouble, so you can get reasonably fine detail with a little work even with a $200 consumer camera. Or, alternatively, hire a professional photographer and have him/her take really high resolution photos of the maps. The advantage of this approach is that you don't have to take the maps anywhere or do anything special with them. Just lay them out on a low table or the floor and align a camera over them, and take heavily-overlapping shots.
Large-format scanners might cost some serious coin even to use for a one-time project like this, but would probably yield better results with less effort.
You might check with local companies that deal in maps and cartography, they might be able to recommend ways of saving the imagery, and some might even offer to help out if the maps may be of commercial interest (they might even share the proceeds with you in addition to giving you high-res digital images).
But I'd say if the maps are truly delicate, your first focus should be to take the highest-resolution images you can of them now, even if it's multiple images per map that need to be stitched. That way, you have *something* preserved in case one or more of the maps is destroyed or deteriorates further before you can preserve it.
If there are particularly interesting features of the map, use the MACRO feature on your camera - most stitching programs can integrate images at different scales and preserve a lot of detail. I used the "Hugin" pano toolkit (free) to stitch together about 100 random photos I took at the top of the Eiffel Tower into an impressive contiguous 360 panoramic shot, and it was literally a "here are the pictures, figure it out" process. The pictures were all taken at different zoom levels, different angles, and all sorts of issues, yet it looked like a Google Street View 360 image. This was 5 years ago, I can't imagine how much better the technology is today.
The geolocation shouldn't be all that hard - it's a matter of choosing a few points on the map and identifying their coordinates accurately. Of course, if there are few/no reference points it gets a lot harder. http://www.openstreetmap.org/ [openstreetmap.org] is a good starting point to a group that does free, open-source mapping. They or some of their related sites might possibly have a tool that does what you want. Also, a professional cartographer may be able to help you out as well.
Re:Hard to Do (Score:1, Interesting)
Where the wrongness is a matter of scales, this can be handled by some deft stretching. You identify points - crossroads are good - by hand, and bend and spindle the scanned map accordingly. When a map is hilariously wrong this still gives more-or-less garbage, but even medieval maps often have the villages (or churches) in roughly the right layout, so stretch-correcting is useful.
The OpenStreetMap folks use this to pimp aerial photography and scanned old maps for use as base layers prior to tracing by mapping minions like me. Try the OSM-Talk mailing list as a point of contact: http://lists.openstreetmap.org/listinfo/talk
Lots of work required...believe me, I know (Score:3, Interesting)
I spent a summer doing this in grad school for the Vegetation Type Mapper [berkeley.edu] project at UC Berkeley. I'm not going to lie to you--it was a ton of work. But the results were cool. The site has all the old maps georeferenced, plus ways to download them.
Needless to say, the library was involved in the project, as was a giant scanner. We relied on ERDAS Imagine software to georeference the old maps to current USGS base maps. There was also a lot of accuracy assessment involved to make sure we minimized error in the georeferencing process. Probably one of the trickiest parts was making sure the old landmark you were using as a control point had not substantially changed in the intervening decades.
My professor and her colleagues published a paper detailing the project [berkeley.edu].
Re:Talk to a curator (Score:3, Interesting)
The geocoding is going to be a BITCH.
The scanning, you'll find out is going to be the easy part.
As a collector of maps, you should know that old maps use different projections to display a 3D object on a 2D surface--the Earth on paper--and that in addition they also use different scales. Sometimes convention is tossed out the window and a map uses neither standard projections nor consistent scale. The older the map, the more this is likely.
In order to apply coordinates to these maps, coordinates that are usable for anything other then simple viewing, you will have to find some way of morphing a grid with coordinates across the images you have after scanning. It might be something as simple as creating a transparent layer in Photoshop that can be stretched to align properly. My guess is that you would need some sort of custom plug-in for this to deal with the various projections used on the map images. Scaling shouldn't be an issue unless it is inconsistent and changes across the image.
Any graphics whizzes out there that can expand on this?
I agree with the posts suggesting photography as a means to capture the images. Glass over the map and careful consideration of reflections on said glass, combined with rather inexpensive camera gear will produce something on par with a desktop scanner. Even cheap cameras these days have pretty high definition. Most definitely the least destructive.
Crowdsource the geocoding (Score:3, Interesting)
I can't speak on how to get them digitalized, but once you do, look to the web to get others to help geocode and get them into shape for overlays. Put them online in a liberal CC license and invite other people to use them. Given the popularity of google maps and the community that's grown up making mashups and apps, I'd be willing to bet there already existing communities of people good at, and interested in, doing this.
Re:Blue print company (Score:5, Interesting)
no macro lens.
What the OP needs is called a large format copy stand.
You want a lens that is "normal" to your film size, thus 50mm on 35mm film as an example.
for an APSc size sensor in a digital camera I think the normal lens would be ~30mm.
I don't know if you want color information or not but honestly either way I would shoot film, then scan the result.
For B&W, shoot Technical Pan film as an ISO of 6, develop in technidol developer (enlist advanced photo class people at a UNI to help with this). For color use Fuji Velvia at 50.
The photo class people will have the copy stands and appropriate cover glass / filters to get the contrast you want as well.
While I have not done maps, I have done large hand drawn artwork this way and the result is vastly better than you would get from directly shooting on a digital (IMHO).
Cheers,
-nB
Dealing with projections talk to an astronomer. (Score:3, Interesting)
We deal with old data with weird and sometimes unidentified projections fairly often. You will need at least some landmarks, though. The larger the scale, the more you will need. The more complex the projection, the more landmarks you'll need. With really old maps that are hand drawn and don't match a distance scale (pre-17th century) you're probably out of luck even if it's a local street map.
Astronomical software to deal with converting between projections is typically open source, but the learning curve is steep. Don't even think about using commercial image editing software. Even if you think it might be doing things right, you'll never be sure unless you're writing your own plugins.
Check a university or major library (Score:1, Interesting)
Either ask the library (if you're in a big city) since they're always interested in getting their hands on anything. Or, as others suggested, find a local university that has a geography or geomatics program - they'll most likely have the expertise and tools to get it done.
As for georeferencing, I've only ever done it with ArcGIS 9, and I don't think you'll want to pay for that, but any self-respecting university will have copies of it on their computer. They may even have historical imagery from the past 60-70 years that'll help you line the stuff up.
Re:dig camera (Score:4, Interesting)
Stitching these together requires 0 effort in any modern photographic editing software.
I actually wrote a little piece of software to automate the stitching process. Just feed the script a bunch of scanned images and it'll align and stitch the images for you. Never got around to make a proper OSS project for it or advertise it, though.
It's here: http://mairas.net/wiki/Mapstitch [mairas.net]
Re:Contact a Museum (Score:3, Interesting)
Re:Blue print company (Score:3, Interesting)
His point is that it won't - It's a discussion that regularly comes up on photography forums. Color accuracy is generally considered better (with the right film) and I've heard resolution figures of 30 or 40 MP for 35mm film. This does assume you scan the negatives properly with a wet drum scanner, an expensive and complex piece of equipment.
I shoot digital myself, but I'm planning to get a hold of a film body to do some comparisons. I don't have a scanner, nor can I develop, so then it becomes a question of how good is my local photo shop.
For the scanning, this is an interesting solution (Score:3, Interesting)
http://www.cs.ubc.ca/~heidrich/Papers/EG.04.pdf [cs.ubc.ca]
"Design of an Inexpensive Very High Resolution Scan Camera System".
500 megapixels for $1600 versus scanning-back cameras that cost tens of thousands. Make the error correction a little bit simpler by rotating the scanning back ninety degrees for four 4-color shots (16 exposures), overlaying them, and taking the median pixel values, at the expense of some resolution.
There are four considerable challenges here, though, not just one, and this project could stall on any of them:
Image capture
Projection determination & georeferencing
Digitizing features & establishing topology
Geocoding