M E D S K Y
MEDSKY is a program for constructing sky flats for direct imaging,
using an algorithm due to Schneider, Schmidt and Gunn.
A list of images is read in from file,and the medians of
equivalent pixels in all the images are found, eg. for N images
(of any size), the first pixel ( the top-left one) in each image
is fetched : the median of these N pixels is then found, the
result becoming the first pixel in the output image. This process
is repeated for all the pixels in the input images to construct a
complete output image. In practice, more than one pixel is
analysed per pass through the image file list, the number being
determined by the memory available to the program. MEDSKY attempts
to minimise the page faulting which would occur by accessing too
much virtual memory at once by processing fewer pixels per pass,
on the assumption that the extra overheads in doing more passes
will be more than offset by fewer page faults. This approach seems
to have been justified, but performance is very sensitive to
changes in the proportions of memory allocated, and it is possible
that improvements could be made.
Command parameters -
FILES The name of a .DAT file containing a list of names
of images. All these images must have dimensions
equal to those of the first image in the list, FIRST.
OUTPUT The name of the result of the operation. This can
be the same as for FIRST. If not, a new structure
is created, with everything but the data a direct
copy of the input.
IMGLOG Only used if the LOG keyword is set. This supplies
the name of the 'log' image created. This will contain
a single integer data array.
Important
~~~~~~~~~
MEDSKY does not check that the output image name is not a member
of the image file list.
Command keywords -
SCALED If set, MEDSKY attempts to compensate for differences
in data scale between the different images. It does this
by conceptually scaling all images so that they have the
same median value as the first image.
LOG If set, MEDSKY creates a 'log' image. This is an
image with the same dimensions as the output image, with
each pixel a number from 1 to N (the number of files)
showing the image number which for each output pixel had
the data value closest to the median value calculated.
This can be used to see if any image dominates the others,
or if there is any trend across the image. This is a rather
specialised option, and is not expected to be used much.
User variables used - None
DJA / AAO. 16th Aug 1987
FIGARO A general data reduction system