Computer Vision Algorithms Used to Solve DARPA De-shredding Challenge
Monday, December 5, 2011 at 9:01AM Last month, DARPA challenged the world's geeks to solve a seemingly unsolvable riddle: How do you reassemble documents that have been shredded into more than 10,000 tiny pieces? Turns out computer vision technology is the answer.
A small San Francisco-based team, called "All Your Shreds Are Belong to Us," correctly reconstructed each of the five challenge documents and solved their associated puzzles. The team used custom-coded computer-vision algorithms to suggest fragment pairings to human assemblers, who would then verify whether the pieces fit.
It was no small feat: the winning team spent nearly 600 man-hours developing algorithms and piecing together the documents.
And what, you're wondering, was the point? DARPA explained the significance of this achievement on its website:
The Shredder Challenge represents a preliminary investigation into the area of information security to identify and assess potential capabilities that could be used by war fighters operating in war zones to more quickly obtain valuable information from confiscated, shredded documents and gain a quantitative understanding of potential vulnerabilities inherent to the shredding of sensitive U.S. National security documents.
So, shredded documents are no longer secure? Great. I wonder what company will be the first to develop the new in-home document burner?
DARPA,
Data security,
computer vision in
computer vision 


