Algorithms To Reassemble Ancient Map of Rome
MattJ writes "The BBC reports on a Stanford project to reconstruct pieces of an ancient, detailed, 60' by 43' 'map of Rome carved into stone slabs about AD 210 but later broken into fragments.' ... So complicated is the jumble of parts that for decades the map pieces have been referred to as 'the biggest jigsaw in the world.' Researchers developed algorithms to assemble the 3D fragments of the map, and have had some success."
Once it's reassembled, we'll be able to frag each other on it?
Sweet.
Hmm, does this qualify as a 'crack', a 'cheat' or a 'trainer' for jigsaws?
Slashdot Sig. version 0.1alpha. Use at your own risk.
to these people's software...
Donald 'Duck' Dunn: We had a band powerful enough to turn goat piss into gasoline.
Don't worry. Just an advance publicity stunt for the new Indiana Jones movie.
Don't blame Durga. I voted for Centauri.
Just do all the outside pieces first, and work your way in.
Neck_of_the_Woods
#/usr/local/surf/glassy/overhead
Is she a prof at stanford now!?
She said, "I said don't worry about it. I'll get one of my students to fix it.", when I broke that vase in her kitchen
Perhaps I'm beeing to pessimistic, but I don't think they'll get to far with this project, unfortunately. There is, after all, only about 15% of the map which has been found, in no less than 1,186 pieces. Many of them probably very similar. As the article says, the program "has found seven high-probability matches and a host of other possibilities".
But the idea is very cool. Probably this could make it easier to put together pieces of other artifats like clay tablets and the likes, and speed up the field of archaeology, which has in some cases been painfully slow.
Cuneiform recognition software, anyone?
The belief in a biblical god is an ignorant one
Similar techniques have been applied to frescoes and estelas of ancient buildings with somewhat higher rate of success. This map seems a particularly hard instance given the similarity of the patterns while at the same time having no global regularity. In contrast building designs often repeat patterns, so you only need to solve the puzzle once for a pattern and thereafter apply the same solution.
In my high school biology class, my teacher made an incredibly insightful comment that I've found applies to many fields of problem solving:
"If it's going to take 10 years to solve the problem using current technology, then spend 3 years working on making new technology, then 3 years using the new tech. Now you've done it it 6 years instead of 10, just by 'procrastinating' and working on your tools first!"
From the BBC article, we see Professor Marc Levoy's quote in highlight, "In doing so, we have created the largest and most detailed model of a cultural artefact"
Think about this for a sec... they've invented tech that provides a better model for cultural artifacts than we've ever had before. Sounds like a key boost in the right direction for this field of engineering.
I thought the title said "Algorithms to Resemble Ancient Maps of Rome".
I thought someone had peeked at my perl code!
*ducks*
Jon Jannotti wrote a technical article on this over at Ars a few years ago, if anyone's interested in more information about the project and its techniques. Sorry for the shameless plug, but what can I say but that I love Rome :)
Nice to see this technique advance.
I have a November 1970 issue of National Geographic with an article entitled "Computer Helps Scholars Re-create an Egyptian Temple" by Ray Winfield Smith that describes something very similar. A summary is embedded here.
The article says that they had to deal with as many as 30,000 pieces, many of which had disappeared over time. They shot B&W pictures of the pieces at the rate of 400/day, and then created punched cards (and then tape image) for each picture. Each card had the piece number and codes for various attributes (such as paint color, figures, hieroglyphs) of the piece. They could then search for other pieces that matched attributes of a subject piece. IBM supplied them access to a computer.
BTW, the issue's headline article was "Behold the Computer Revolution", a nice historical piece now.
Rome wasn't built in a day...
16000 sacks of shredded paper and 600 million individual scraps of paper seems a bit more demanding task.
Probably most of the same algorithms would apply to both problems.