itu-t81 part 2 (jpeg compliance testing)

[Nils]

Hi,

I am writing a jpeg decoder based on standard itu-t81. I can find links
to itu-t81.pdf everywhere, but now I would also like to obtain the part
2 mentioned in this one, so I can test my decoder for compliance. I
have googled around but can't find it.

Does anybody know where to find it?

Thanks in advance,

Nils Haeck
www.simdesign.nl

[Thomas Richter]

Nils wrote:
> Hi,
>
> I am writing a jpeg decoder based on standard itu-t81. I can find links
> to itu-t81.pdf everywhere, but now I would also like to obtain the part
> 2 mentioned in this one, so I can test my decoder for compliance. I
> have googled around but can't find it.
>
> Does anybody know where to find it?

Sure. It's an ITU and ISO standard, so you can buy them from the ITU or
the ISO. URLs for the ITU and the ISO are obvious. (-: "Free copies" you
find in the net are likely illegal, or cover only the FDIS state of the
standard, i.e. may contain errors that have been addressed in later
corrigenda.

Furthermore, I would also suggest to check the following book:

William B. Pennebaker, Joan L. Mitchell:
JPEG Still Image Data Compression Standard
Van Nostrand Reinhold, New York, ISBN 0-442-01272-1

Aka, "The Pink Book". (Yes, it's really pink.)

So long,
Thomas

[Nils]

> Sure. It's an ITU and ISO standard, so you can buy them from the ITU or
> the ISO. URLs for the ITU and the ISO are obvious. (-: "Free copies" you
> find in the net are likely illegal, or cover only the FDIS state of the
> standard, i.e. may contain errors that have been addressed in later
> corrigenda.

Well I obtained part 1 from the W3C organisation website, so unlikely
to be illegal :)

Thanks for the info. I guess I'll have to buy the part 2 then.

> Furthermore, I would also suggest to check the following book:
>
> William B. Pennebaker, Joan L. Mitchell:
> JPEG Still Image Data Compression Standard
> Van Nostrand Reinhold, New York, ISBN 0-442-01272-1
>
> Aka, "The Pink Book". (Yes, it's really pink.)

Thanks for the tip, but I already know a lot about the compression
principles used in Jpeg, and it's a bit pricy at US$155. Btw I'm not
very interested in the arithmetic coding and hierarchical compression
parts of Jpeg, I just want a decompressor that handles about 99% of all
jpegs available (huffmann sequential and progressive, IDCT). I guess
things like using MMX and SSE2 for the IDCT stage are not in the book
anyway :)

Kind regards,

Nils

[Pete Fraser]

"Thomas Richter" <th...@math.TU-Berlin.DE> wrote in message
news:4jb6aeF...@news.dfncis.de...
> Nils wrote:

>> Does anybody know where to find [T.81 part 2]?

>
> Sure. It's an ITU and ISO standard, so you can buy them from the ITU or
> the ISO. URLs for the ITU and the ISO are obvious. (-: "Free copies" you
> find in the net are likely illegal, or cover only the FDIS state of the
> standard, i.e. may contain errors that have been addressed in later
> corrigenda.

I think you're looking for ITU T.83, a.k.a. ISO10918-2.

You used to be able to sign up with ITU once a year to download up to three
standards a year for free. I don't know if that still works.

[Thomas Richter]

Pete Fraser wrote:

>>>Does anybody know where to find [T.81 part 2]?
>>
>>Sure. It's an ITU and ISO standard, so you can buy them from the ITU or
>>the ISO. URLs for the ITU and the ISO are obvious. (-: "Free copies" you
>>find in the net are likely illegal, or cover only the FDIS state of the
>>standard, i.e. may contain errors that have been addressed in later
>>corrigenda.
>
>
> I think you're looking for ITU T.83, a.k.a. ISO10918-2.
>
> You used to be able to sign up with ITU once a year to download up to three
> standards a year for free. I don't know if that still works.

The second alternative is to become a member of a standardization group.
If you're working for a university, the membership is, at least for
Germany and the DIN, for free.

So long,
Thomas

[snork...@gmail.com]

>
> Thanks for the tip, but I already know a lot about the compression
> principles used in Jpeg, and it's a bit pricy at US$155. Btw I'm not
> very interested in the arithmetic coding and hierarchical compression
> parts of Jpeg, I just want a decompressor that handles about 99% of all
> jpegs available (huffmann sequential and progressive, IDCT). I guess
> things like using MMX and SSE2 for the IDCT stage are not in the book
> anyway :)

I suggest you take a visit to your nearest University library. They
will generally have some ISO and ITU documents on hand, and probably
have a subscription that will allow them to access to the online
archives. A reference librarian can probably hook you up with the
appropriate file in minutes. (Different libraries will have different
methods for protecting the copyrighted content - you may be restricited
to viewing the document on a locked-down PC.)

It is a shame that the standards bodies charge so much for these
documents, as it locks out ordinary citizens from a lot of access. But
until you come up with an alternative method of funding their activies,
I'm afraid we are stuck with the status quo.

- Mark Nelson - ma...@ieee.org

[Guido Vollbeding]

Nils wrote:
>
> Well I obtained part 1 from the W3C organisation website, so unlikely
> to be illegal :)

The JPEG standard text (10918-1/T.81) is really in the public domain now,
it is available from various places for free.

> Thanks for the info. I guess I'll have to buy the part 2 then.

This part 2 ("compliance testing") is not really necessary.
You are probably wasting your time to dig this thing up.

Anyway, there is an opportunity on the ITU site to sign up a registration
for downloading up to three documents per year for free.
It is a bit difficult to find, but then I did this and it worked.

> > Furthermore, I would also suggest to check the following book:
> >
> > William B. Pennebaker, Joan L. Mitchell:
> > JPEG Still Image Data Compression Standard
> > Van Nostrand Reinhold, New York, ISBN 0-442-01272-1
> >
> > Aka, "The Pink Book". (Yes, it's really pink.)
>
> Thanks for the tip, but I already know a lot about the compression
> principles used in Jpeg, and it's a bit pricy at US$155.

This book is still the best reference for JPEG in existence today
(well, until I write my own book...:) and highly recommended for
anyone doing serious JPEG work.

> Btw I'm not very interested in the arithmetic coding and hierarchical
> compression parts of Jpeg, I just want a decompressor that handles
> about 99% of all jpegs available (huffmann sequential and progressive,
> IDCT). I guess things like using MMX and SSE2 for the IDCT stage are
> not in the book anyway :)

The DCT/IDCT part is in fact the weakest point in this book, unfortunately.
There are no useful algorithms given for practical implementation, and it
lacks explanation of the fundamental DCT property for image representation.
That is the reason that all subsequent image coding research went wrong.
And that is the reason that I have to write my own book.
Joan Mitchell has already declared to be editor of my book (and contributing
some content...:).
By the way, the "Pink Book" is no longer available in the "Van Nostrand
Reinhold" edition - it is now available in a "Springer Publishing" edition,
as Joan Mitchell told me at the Geneva meeting this year (the new book will
also be published there).

Regards
Guido Vollbeding
Organizer Independent JPEG Group

[Thomas Richter]

Nils wrote:

>>Furthermore, I would also suggest to check the following book:
>>
>>William B. Pennebaker, Joan L. Mitchell:
>>JPEG Still Image Data Compression Standard
>>Van Nostrand Reinhold, New York, ISBN 0-442-01272-1
>>
>>Aka, "The Pink Book". (Yes, it's really pink.)
>
>
> Thanks for the tip, but I already know a lot about the compression
> principles used in Jpeg, and it's a bit pricy at US$155.

Who said about buying? (-: The nearest university to your place will
likely have it. Check in the computer science, electrical engineering or
communications engineering department.

> Btw I'm not
> very interested in the arithmetic coding and hierarchical compression
> parts of Jpeg, I just want a decompressor that handles about 99% of all
> jpegs available (huffmann sequential and progressive, IDCT). I guess
> things like using MMX and SSE2 for the IDCT stage are not in the book
> anyway :)

Part-2 might be very handy, but in reality you'll find a couple of JPEGs
around that do not really follow the standard anyhow, so that part might
be not as useful as you might think immediately. First of all, 99% of
the JPEGs out there only use a small percentage of the features, but
it's not too unlikely that you'll find some that do not follow the
standard but you would like to get decoded. For that, collect experience
by experimenting, or (probably a faster approach) get in touch with
people that do have this experience.

So long,
Thomas