Usenet posting on Acrobat 6 JPEG 2000 compresion - poor results with grayscale images

From: jfaughnan@spamcop.net (John Faughnan)
Newsgroups: adobe.acrobat.windows
Subject: JPEG2000 and grayscale image size growth
NNTP-Posting-Host: 208.138.188.194
Message-ID: <5c0dbfb4.0404081405.1e8f3bd@posting.google.com>

I did some initial testing using the most extreme (low quality) JPEG
2000 image compression settings with Distiller 6. (JPEG2000 is new in
Acrobat 6. A primary potential application is scanning color documents
including maps.)

I first scanned a sample document at 200x200 16 bit color, producing a

10MB tiff (lzw compressed) file. The JPEG 2000 PDF of this image was
only 160K. The text in the 160K file was quite readable. I considered
this to be a very good result, almost a 70 fold compression with
preserved text readability. In my past experience JPEG compression of
a scanned text image makes the text unreadable due to jpeg artifact
even with moderate (10 fold) compression. This is a qualitative
improvement over JPEG. (For reference, past experience using B/W
images scanned with CCITT 4 compression produces typically a 40K image
of the same test document.)

I then scanned the same document at 200x200 gray scale. This resulted
in a 3.6MB TIFF (lossless compression). The JPEG2000 compressed PDF,
however, was 1.77MB! A JPEG PDF of the same file was only 300K (and
was quite readable). Something's wrong here - I expected a JPEG2000
maximally compessed PDF of this grayscale image to come in at about
60-80K. I suspect a bug in Distiller's handling of JPEG2000 compressed
grayscale images. I wonder if Distiller is not honoring the
compression setting for grayscale images.

Has anyone seen anything like this?

john
jfaughnan@spamcop.net

meta: jfaughnan, jgfaughnan, jpeg2000, jpeg 2000, pdf, jp2k, lossy
compression, scans, scanner, scanned image, acrobat, adobe acrobat 6,
text