NOTE: I found that we already have the HCOMPRESS stuff unpacked in /u3/extra
on tucana. The paper this is based on that Dave mentions is in the doc
subdirectory.
>From harrisd@ll.mit.edu Tue Feb 1 09:47:20 1994
To: fitz@noao.edu
Subject: Re: SAOimage display problems for certain images Mike, Thanks for the pointers to the STSDAS and XRAY
information...I'll check them out...and for the information
about the other packages. Here's some information about Rick White's software: The purpose for which he developed his image compression
software was for the "Compressed Digitized Sky Survey on
CD-ROMs", some of which is now available for purchase from
the Astronomical Society of the Pacific. This project at
STSCI used a laser scanner to scan and digitized about 1,500
photographic plates from the "southern SERC J band survey"
and the "northern Palomar E-band survey" to produce (I
think) 1,500 14,000 x 14,000 16-bit/pixel digital images. Obviously, that's quite a bit more data than most people's
systems can handle, so part of the project involved
developing an image compression algorithm to convert
smaller-sized raw images into compressed versions that would
still retain reasonable astrometric and photometric quality.
Sets of those images at two different levels of compression
would be reproduced on CDROMs and sold (the sets aren't
cheap: the price is $2,900 for orders before 15 February,
but $3,500 after that...so be sure to get your check in the
mail soon, Mike
The problem was compressing typical astronomical images from
photographic plates, CCDs, etc., is that they contain lots
of localized, point sources, which implies high spatial
frequency content. There are lots of popular data and image
compression methods, but none of them are very well suited
for such imaged. For example, the JPEG compression scheme,
which is very popular now, uses the Discrete Cosine
Transform on small blocks of the image and it achieves
compression large by throwing away the coefficients of the
high frequency terms. Thus, using JPEG for astronomical
images would introduce undesirable artifacts when images
with point sources undergo lossy compression and then
decompression to a viewable or processable image. So, White developed an algorithm based on a two-dimensional
Haar transform (a sort of wavelet transform). I don't think
the idea is his, since his paper cites work 10 to 15 years
ago along the same lines. But, you can get a copy of his
software that runs on Unix or VMS systems and try it! Here's where to get his software and its documentation: ftp to stsci.edu and usual anonymous ftp login
procedure change to the directory /software/hcompress In it you'll find these files: OWNER (text)
aareadme (text)
backup.com (text, for VAX)
hcompress.bck (binary, VAX Backup file)
hcompress.tar.Z (binary, Unix tar.Z file)
modify.exe (binary, VAX program) The backup.com program unpacks the files in hcompress.bck
on a VAX. The modify.exe program (if my memory is correct)
is needed to adjust the record lengths of some files on
VAX/VMS systems. I installed his software last summer on
a VAX here, but could do much with it since I had no display
capability and no CCD images to work with. But, the
installation was easy and the code worked. The Unix stuff is in the .tar.Z file. It, too, is easy to
install and I did it on several Unix systems that could run
the code, but with no display capabilities, until we got
our SPARC. In fact, I think my boss back at the main Lab
was impressed enough with White's algorithm and his test
cases that he was able to get the money for the SPARC for
me to develop code starting with White's code. The documentation is in OWNER, aareadme, and in files
inside the VAX and Unix archive files. It consists of
instructions on how to install the software on either
kind of system, Unix "man" pages for the executable
commands (hcompress, hdecompress, fcompress, and
fdecompress), and a PostScript file containing his
paper: "High-Performance Compression of Astronomical Images" I don't remember if the paper was published in a journal
or not. The paper contains the older references that
I mentioned above, and I've managed to accumulate most
of them. If you want the exact titles, etc., for any
of them, just let me know. There is also a related paper: "Efficient Transfer of Images over Networks"
J. W. Percival and R. L. White
that was published in "Astronomical Data Analsis
Software and Systems II", which is Volume 52 in the
ASP's Conference Series. Percival is the fellow at the Univ. of Wisconsin's Space
Astronomy Laboratory who has been integrating White's
software into a TCP/IP-based communications protocol
that will be used for the WIYN telescope and, it appears,
for others as well. I've talked to Jeff Percival about his
work and he has been very helpful, even providing to me a
copy of his code that I have installed on the Sun SPARC. Finally, here are their email addresses, in case you need
them: White: rlw@stsci.edu Percival: jwp@bernie.sal.wisc.edu I found both of them to be very helpful, though you may
not get an quick response from Rick White, since he's
been very busy for the past few months, probably, because
he has been involved with the pre-fix HST image restoration
work and with the preparations for and testing of the
HST hardware changes. Dave Harris
>From fitz Tue Feb 1 10:45:04 1994
To: harrisd@ll.mit.edu
Subject: Re: SAOimage display problems for certain imagesHi Dave,
Thanks for the information, I'm sure it will come in handy. Turns
out that on an obscure part of our disk we already have the software, I
just didn't know about it. Your notes also help explain things more clearly.
I only hear things occassionally about remote observing at KPNO, the
person to contact would be Rob Seaman (seaman@noao.edu) who is the IRAF pro-
grammer responsible for the ICE (IRAF Control Environment, an IRAF based
data aquisition system) on the mountain. He's worked with observers who
have tried remotely working on the mountain and would know how well that
has worked in the past.Cheers,
-Mike