SourceXtractor performs the detection on a single, coadded, image, which can be created using, for instance, SWarp.
Later, position and shape can be projected into the individual measurement frames whenever required. For more details on how does this work, have a look at the measurement example.
In this particular example - see detection.conf -,
SourceXtractor will perform the detections over the coadded image
sim11.fits, using the weight map sim11.weight.fits,
and writing the output into detection_cat.fits.
The settings used will be, mostly, the defaults, except for some parameters worth mentioning:
partition-multithreshold=trueenables the partition of sources that are too close together, and are detected as one.grouping-algorithm=SPLITtells SourceXtractor to group sources that were split previously. In this example, this is not particularly useful, but it becomes handy for model fitting, as it allows to fit multiple close sources at once, so the effect they each have in one another is taken into account.use-cleaning=Trueenables the removal of spurious sources detected on the edges of bright ones. It does so via a simpler model fitting, using a Moffat profile.output-propertiesSourceXtractor works on properties, not on columns. Each property may have on or more attributes that end being written into columns. You can check the supported properties viasourcextractor++ --list-output-properties. You can also have a look at the actual columns withsourcextractor++ --property-column-mapping-allcheck-image-*The software can also generate FITS image useful for debugging itself, or the configuration. In this particular case, we are asking for images with the interpolated background, the variance map, the segmentation map (before partitioning and cleaning), the partition map (after partition and cleaning), and the grouping. These images include the WCS headers extracted from the underlying image ( the coadded detection one, in this case)
To run this example:
sourcextractor++ --config-file detection.conf