[ImageDM] Derived objecta

Douglas Tody dtody at nrao.edu
Tue Nov 19 13:06:29 PST 2013


There are a few words on this in the ImageDM draft:

     Derived metadata is metadata pertaining to the observed target that
     is derived by analysis of the current dataset.
       ...

     For many image datasets derived metadata is not relevant and may be
     omitted, but if a central target is analyzed derived attributes may
     be specified.  For example, a cutout or high-resolution spectral or
     time cube with a specific target might have derived attributes for
     the target such as redshift or variability amplitude.

While it is true that for much image data the Derived element may not be
useful, there are use-cases where it can be, just as for spectra or time
series.  A Spectrum or TimeSeries is usually an observation of a single
object, for which these Derived values are clearly useful.  But an Image
can be a high resolution image or cutout of a single object as well, and
all the same logic applies.  There may well be other cases as well,
where we derive quantities that pertain to the overall image, e.g.,
statistical measures.

So while it may not be used all that much for Image data, there are
cases where it is still needed.

 	- Doug




On Tue, 19 Nov 2013, CresitelloDittmar, Mark wrote:

> Small thread..
> The imageDM contains a 'Derived' object which is carried over from the
> Spectral model.  This object definition:
> 
> Derived:
>     - SNR
>     - Redshift.*
>     - VarAmpl
> 
> I'd like to clarify the 'domain' for this information.  It seems rather
> 'Spectral" related, so I'm not sure if it needs to be included in the
> generic ImageDM at all, but can see that certain types of images might
> want this sort of information.
> 
> Mark
> 
> 
>


More information about the dm mailing list