JPEG Patents
Date: Thu, 12 Jan 95 18:38:37 EST
From: "Joan Mitchell (8-863-7238)"
To: Tim Berners-Lee <timbl@w3.org>
As one of three editors of the JPEG Part 1 document, I am responding
to your inquiry. The "Requirements and Guidelines" document, called
ITU-T Rec. T.81 | ISO 10918:1, became an official CCITT
recommendation in Sept. 1992 and an ISO published standard in
February 1994. Except for the foreword, the text is common. Part 2
"Compliance Testing" is probably also an official standard now. Part
3 "Enhancements" is working its way through the Draft International
Standard (DIS) ballot.
The IS version is for sale from your national standards body. In the
US the number to call at ANSI is 212-642-4900. Bill Pennebaker and I
got permission to publish a copy of the Draft International Standard
Part 1 and 2 in our book "JPEG: Still Image Compression Standard",
Van Nostrand Reinhold:New York c. 1993 ISBN 0-442-01272-1. It sells
for $59.95 in the US. Anyone who is seriously implementing will need
an official copy of the IS rather than the DIS.
The JPEG international committee tried hard to have early disclosure
of patents. So far no one has pressed any claims on the JPEG baseline
(sequential Huffman) system that I know of. I can not take responsibility
for knowing the future however. I have not attended the JPEG meetings
for nine months now so I am a bit out of date.
I have appended below
Annex L from the Draft International Standard
10918 Part 1 that discusses patents. Below it are the names that
were included in the final International Standard. In a separate
e-mail, I'll send you the text of IBM's letter relative to licensing
arithmetic coding patents to practice standards. John Lowe is the
official person to contact for a license from IBM. The JPEG/JBIG
arithmetic coding patent holders (IBM, AT&T, and Mitsubishi Electric
Corp.) have all issued statements indicating a special deal for
implementing those standards. I would recommend that you contact
them directly. You also may want to contact Eric Hamilton, Chair of
JPEG, (eric@c-cube.com) for more information. Tom Lane, Independent
JPEG Group, may also have lots to say on this topic.
JPEG is not appropriate for color images after they have been
converted into 256 palette values. So you may want to look at
other standards. For example, during my leave of absence to write
the MPEG book, I am working part time at IBM Burlington, VT. The
group here makes chips (LZ-1) that run between 5 and 40
Mbytes/second. The encoder certainly has proprietary things in it.
But to the best of my knowledge, decoders would not be so likely to
hit intellectual property. The algorithm, called ADLC, is right now
winding its way through the European Computer Manufacturers Assoc.
standards program and the Quarter Inch tape Cartridge standards. If
you want more information, Bill Lawrence (802-769-6685) would be
happy to supply anyone with some literature and even a diskette that
runs the algorithm so its compression can be tested.
Sincerely,
Joan Mitchell
P.S. I expect that you will want to propagate this answer. You have
my permission.
Appended