Re:The Kopime Vector
by mistercow on Tuesday May 01, 2012 @12:16PM
(#3915056)
You could narrow down the possible matches to just a handful by comparing movie lengths – say, within 10 seconds of the same length is a possible match. Maybe throw in some basic title comparison to narrow the search further, and then do a simple check to see if the audio tracks seem to line up (an exhaustive comparison isn't necessary) to rapidly detect and exclude duplicates with a more than adequate accuracy rate. It would require a bit more complexity in the Kopime device, but not so much as to make it impractical.
[ Parent
| on HN ]
Re:The Kopime Vector
by pessimizer on Tuesday May 01, 2012 @12:49PM
(#3915247)
Are you aware of any (free or open) code out there that has implemented a reasonably efficient fingerprinting system? I'd love to take a look at it and/or use it for a project I'm working on.
I don't know how efficient it is, but here's an opensource perceptual hash library: http://www.phash.org/
Re:The Kopime Vector
by canadaduane on Tuesday May 01, 2012 @02:05PM
(#3915630)
Very interesting. Thanks!
Re:The Kopime Vector
by pessimizer on Tuesday May 01, 2012 @02:52PM
(#3915862)
New to me. Thanks a lot.