RE: Data compression

["andreas axelsson" <andreas.axelsson@dynamics.saab.se>]

Hi,
I've now looked through the TN-55 and also taken part of your comments. I
still think it could be
valuable to do some work on the compression. Especially since I find it
quite interesting! :-)

Arne mentioned gzip as a comparsion and that it yielded a compression of
approx. 30 %. Keeping
in mind that gzip doesn't take advantage of any structural information (not
even the fact that
it is 16 bit pixels) I expect one can achieve far better lossless
compression with a good model
and a fast entropy coder such as an arithmetic coder.
Also if the structure of the compressor is designed well one can probably
get a seemless transition
between lossless and lossy compression using the same front-end/back-end.

Could anyone send me a few typical FITS files that I can start take a look
at, or point me to
where I can find them?

Best regards,
Andreas