Re: MIT TOC SEMINAR--ADI SHAMIR--MONDAY--MAY 16--4:15pm

[norm@netcom.com (Norman Hardy)]

At 10:07 5/11/94 -0700, Paul E. Baclace wrote:
>I'm very curious as to how humans can directly decode encrypted
>pictures.  Do they stare at it for 10 minutes and go "ah, there
>it is". 
...
About 10  years ago there was a Scientific American article about visual
encypherment. The decoder required no computing hardware. A one time pad
was available at both ends in the form of an array of 1000 by 1000 random
black or white pixels in the form of a transparency. When it was time to
code a black and white image an array of pixels were produced with each
pixel being black with a probility proportional to the darkness at that
point of the 'plain-image'. That was exclusive ored with the one time pad.
This yielded a random set of black and white pixels and was transmitted
physically by insecure courrier. It it reached the destination it would
ideally be exclusive ored with the other copy of the one time pad. The
receiver could more easily align the cypher-image with the one time pad and
see a fairly good image. This yields the 'and' function in place of the
'xor' and provides about half of the image quality in the information
theortic sense.