[LINK] Airlink

[jim birch]

If the application ran something like a Fourier-transform on a photo you
could get a reliable signature which could then you could do a closest
match to the current library to allow for a bit of smudge or wrinkle. They
are talking about 6 pics a day which is a library couple of thousand items
if retained for a year.  The pic could be tested for uniqueness at
prepublication then recropped or tweaked if its signature happened to
fuzzy-collide with an existing item.  You might have a problem with
straight mug shots but anything else would be ok.

If services like TinEye can find the reuse of images across the entire net
- including altered images with recrops, tints, etc - I can't see that
matching a handful of pics in a paper would be a problem at all.  The main
problem I see is that Airlink might make you wonder why you are reading a
paper at all.  The paper would want some hot and/or well selected
background content to the Internet.

http://www.tineye.com/
http://arstechnica.com/uncategorized/2008/08/tineye-image-search-helps-ferret-out-copyright-ripoffs/


Then, the next step will be when you point your phone at any image at all
and get a Google knowledge graph for it, taking the newspaper out of the
loop.

Jim