Next: User interface software
Up: DESIGN
Previous: Ingest/curation software
  Contents
A major consideration here is in balancing currently
available astronomical data analysis packages with in house software
components.
Given the large volume of data that will be involved, large scale
image and data analysis software will be written using a low level
language. C++ is the language choice because it is object oriented and
allows encodeing of very complex data structures.
Software modules include:
- Utilities for basic image manipulation operations such as pixel
rebinning, resampling to new coordinate systems, etc. Image classes
will be developed in C++ with appropriate I/O interfaces (based on
CFITSIO in the case of FITS images).
- Source extraction, measurement, and classification tools. Existing
software packages include DOPHOT, DAOPHOT, and SExtractor. It would
probably also be necessary to develop bespoke software to provide
additional functionality - an example here would be in the
construction of source catalogues merged across the different
passbands.
- Astrometric and photometric calibration tools. Probably requires a
combination of existing and bespoke software.
- Difference imaging tools. These would be useful for
constructing emission line images derived from subtracting a broad
band (e.g. K) image from a narrow band image (e.g. H2). Also useful for
variability studies. We have software that can generate difference
images by optimally matching the seeing on the two images involved (see
http://www-wfau.roe.ac.uk/sss/halpha).
- We also plan to implement software from drawing upon our
astroinformatics research. Software has been developed that allows
the identification of artifacts such as satellite trails and
diffraction halos in CCD images (Storkey et al. 2002, in preparation)
- Image combining software. Tools for combining CCD frames taken in
overlapping regions of the sky to produce single layer image mosaics
and multiple layered images derived from stacking frames. This would
involve development of special classes to deal with mosaic cameras
such as WFCAM.
As far as on-the-fly applications are concerned,
the feasibility of manipulation of large images is questionable when they
reach Gbytes in size. All of the formal science requirements will be subject
to the limitations of implementation constraints and network bandwidth.
However, we suggest that no user would need or expect rapid response to use
cases requiring manipulation of large imaging datasets.
Next: User interface software
Up: DESIGN
Previous: Ingest/curation software
  Contents
Nigel Hambly
2002-10-08