| Forum Home | ||||
| PC World Chat | ||||
| Thread ID: 121578 | 2011-11-01 11:08:00 | Erayd this sounds like a good challenge. | mikebartnz (21) | PC World Chat |
| Post ID | Timestamp | Content | User | ||
| 1241254 | 2011-11-01 11:08:00 | Hi there all The Darpa Shredder challenge (http://www.shredderchallenge.com/) sounds interesting to me and I reckon it would be good to get a team together for it. Quite a while ago I was looking at creating a program to try and clean up some old documents but it was all a bit mind boggling with the text running in two different directions but this has distinct possibilities. A guy that is well up with computers (has worked for IBM in the states etc) said there could be a great deal of money for someone that could create a program to clean up old documents. |
mikebartnz (21) | ||
| 1241255 | 2011-11-01 11:25:00 | Just checked it out further and you have to be a yank to register so we would need surferjoe on board. | mikebartnz (21) | ||
| 1241256 | 2011-11-02 19:41:00 | Definitely looks interesting, although to be honest I don't have the time for something like that at the moment - I'm rather busy. If you want to take a crack at it though, I wish you all the best :D. |
Erayd (23) | ||
| 1241257 | 2011-11-02 22:13:00 | Just checked it out further and you have to be a yank to register so we would need surferjoe on board. not heard of proxies? :) |
bevy121 (117) | ||
| 1241258 | 2011-11-03 00:55:00 | not heard of proxies? :)The issue would be getting paid if you win... | Erayd (23) | ||
| 1241259 | 2011-11-03 06:22:00 | The issue would be getting paid if you win... +1 |
mikebartnz (21) | ||
| 1241260 | 2011-11-03 21:06:00 | That does look interesting. My first thought for an algorithm would be taking each piece and matching it to each other one side by side so that they are nearly touching (not necessary to literally move them, but helps visualise what I'm talking about), rotate one and measure the minimum variance in gap between them as it is rotated. By this I mean that if we have two pieces that were joined, and they have been cut with an imperfect edge (as is the case in the first image) then if we put them side by side, we can rotate one so that the gap between the pieces is the same at all points. If the gap is uniform to a specified tolerance, it's likely that they are connected. On the first image that'd take a maximum of around 100,000 iterations (in reality less). I'm not going to download the rest but I suspect the others are harder, with perhaps more uniform edges which might render my technique more difficult or impossible. |
george12 (7) | ||
| 1 | |||||