Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!decwrl!deccrl!bloom-beacon!bloom-picayune.mit.edu!athena.mit.edu!pshuang
From: pshuang@athena.mit.edu (Ping-Shun Huang)
Newsgroups: comp.compression
Subject: JPEG compression
Keywords: JPEG image compression query
Message-ID: <1991May26.231855.20247@athena.mit.edu>
Date: 26 May 91 23:18:55 GMT
Sender: news@athena.mit.edu (News system)
Organization: Massachusetts Institute of Technology
Lines: 17
Bcc: pshuang@athena.mit.edu

I was wondering if anyone here knew of any articles in journals/magazines or
would be able to give a brief but lucid explanation of the compression algorithm
being used by the JPEG standard.  I've recently acquired a program for the IBM-PC
called Image Alchemy which knows how to perform JPEG compression and compresses
typically at 5:1 ratios for the sample .GIF images which I have fed into it (i.e.
the JPEG output is 20% the size of the already-LZW-compressed bitmap).  I am
quite curious as to how it is doing this.  What I am hoping for is not exact
pseudo-code but an idea of what the algorithm tries to take advantage of in
image-type data.  The degradation in the images is just barely noticeable but only upon careful observation.

An interesting observation about JPEG (or at least the particular implementation of it which I was using) was that it is the only compression algorithm I know of where decompression takes more CPU time than the compression did.  This may have been a fluke, but it was true for all images I tried and the difference was a factor of two and not just a few percent.

I am also curious about arithmetic encoding, about which I have not heard very much before joining this newsgroup.

--
Singing off,
UNIX:/etc/ping instantiated (Ping Huang).