Next: The Classification
Up: Procedures to Follow
Previous: Helpful Hints
It may happen that program execution is aborted or that the output is
clearly wrong. In most cases it is caused by setting wrong values for
keywords. Check carefully whether keyword values look reasonable. If
there is no obvious mistake made in setting the keywords, run the ANALYSE/INV program again using the DEBUG option. This can be
done by using DEBUG as one of the parameters of the ANALYSE/INV command line. A lot of information will be displayed for
each object. It is possible to follow the program flow and to see how
the cleaning of blended images is done. Because of the large amount of
information displayed, using the DEBUG option is feasible only
for very small frames. Identify on the image display the region which
gives most of the trouble such as not detecting obvious objects,
failure to resolve pairs, or multiple detections of single objects,
and use the EXTRACT command to create a subframe of dimensions
between
and
pixels with up to
objects. Then use the command SEARCH/INV and
finally ANALYSE/INV with the DEBUG option and try to find
out, based on the displayed data, what the source of the trouble is.
Displaying data concerning a particular object starts with its
identification number, and its pixel coordinates. Next comes three
kinds of information, each consisting of 80 values. They are arranged
in eight columns. The first presents a profile in eight octants. Each
column corresponds to a particular octant, starting with an octant
pointing to the right and going counter clockwise. The first row gives
the value of the central point. The second kind of 80-values array
uses only four first columns. The first column gives average profile,
the next three give the first three amplitudes of Fourier expansion of
the objects profile over the octants. In both these cases data has
been divided by background and multiplied by 1000 for easier
management of displays. The third kind of an array contains only 0
and 1. This array indicates how the object was cleaned. The value of 0
indicates that the actual data is used. The value of 1 means that the
data is interpolated. The arrangement of columns and rows is the same
as in the case of displaying the profile. Only the row corresponding
to the central point is missing.
After intermediate data is displayed for all objects, the final
results are shown. They are presented in three separate chunks of data
for each object. The first gives ID, x-coordinate,
y-coordinate, distance to nearest other object, extent of
unblended central part, approximate radius of an object, and
radius of saturated part of an image, all except ID expressed in
pixels.
The second gives data that is stored in an output table. A list of
table column numbers and labels can be used for identifying displayed
quantities. Data from columns :IDENT, :NR_OTH_OBJ, and :AR
is not presented here, as it has been previously displayed.
The third gives first 21 points of one-dimensional profile.
Next: The Classification
Up: Procedures to Follow
Previous: Helpful Hints
Petra Nass
1999-06-15