1993-12-24 - Re: picture signatures

Header Data

From: peb@PROCASE.COM (Paul Baclace)
To: wex@media.mit.edu
Message Hash: dd00605377b17f880d14d2df77722a014e7d3a28df9ffcd1659132b201d3779d
Message ID: <9312240009.AA26174@ada.procase.com>
Reply To: N/A
UTC Datetime: 1993-12-24 00:09:48 UTC
Raw Date: Thu, 23 Dec 93 16:09:48 PST

Raw message

From: peb@PROCASE.COM (Paul Baclace)
Date: Thu, 23 Dec 93 16:09:48 PST
To: wex@media.mit.edu
Subject: Re: picture signatures
Message-ID: <9312240009.AA26174@ada.procase.com>
MIME-Version: 1.0
Content-Type: text/plain



There is no need to sign the picture itself: a secure time stamp would
prove ownership of the oldest version of a picture:

	CryptoHashA(picture) --> H_i 

Then the timestamp service:

	CryptoHashB(H_i, H_j) --> H_receipt ; H_j is hash from another customer

The H_receipt is published in a well-known place/time and is traceable to the 
picture.  The actual scheme the Bellcore folks worked out is tree
structured for efficiency reasons.

An altered version of a photo would presumably have a different hash,
but it would not have an earlier timestamp hash.  Insofar as one could
recognize that an image was incorporated into another altered image,
misappropriated and stolen copyrights could be detected.  Recognition
is inherent in this process. This isn't very different from sound
sampling: if the alteration completely obscures the orignal beyond
recognition, then there isn't much of a case to prove theft.  For a
picture example, I could digitize a copyrighted image and use *only*
it's color palette on one of my pictures (I haven't tried this yet, but
it sounds like a good "empircal color theory").

BTW, Anyone have an email addr of the Bellcore folks who invented the 
secure timestamp system?  I'd like to use the service if it is still 
operational.


Paul E. Baclace
peb@procase.com






Thread