FITS compression

[William Pence]

Lucio Chiappetti wrote:
> 
> > extension.  The compressed image bytes are then stored in a binary table,
> 
> ...being the binary table format already overloaded with different (semantic)
> applications, and being the "variable format" the most awkward feature in it,
> would it not be neater to define "compressed images" as an altogether *NEW*
> FITS extension ?
> 
> Its data area could then be arranged as a stream of compressed data according
> to any of the proposed formats, and just header keywords used to describe it.

We need the full flexibility offered by binary tables for this compression
convention, so we don't want to invent a new type of extension with all the same
features.  In the general case, the binary table containing the compressed image
will contain multiple rows (each compressed image tile goes into a separate
row), and multiple columns.  There may be 2 variable length columns, one for the
tiles that could be compressed and one for any tiles that could not be
compressed.  There can also be any number of other columns in the table
containing other parameters that are needed to compress or uncompress the image
tiles.

Only in the simplest case where an entire image is compressed as a single tile
would you end up with a binary table that has just a single row and potentially
only a single column.

____________________________________________________________________
Dr. William Pence                          pence at tetra.gsfc.nasa.gov
NASA/GSFC Code 662         HEASARC         +1-301-286-4599 (voice)     
Greenbelt MD 20771                         +1-301-286-1684 (fax)