Classification of images

The classification process (command CLASSIFY/IMAGE) tags each frame with one or more character strings supplied by the user that classify the image in a unique way and will be stored into columns of the OST. For instance scientific exposures observed in SUSI using the filter number 639 may be described` with the two strings ``SCI" and ``FILT_639". The practical problem is however that the very large number of different possible optical elements mounted on instruments like EMMI/SUSI is not fixed in advance; for example, filters may be added or changed. Clearly, one do not want to write classification rules for all filter numbers or all combinations of grisms and gratings.
We have for those reasons introduced the concept of `` Wild Card Replacement Character"; This character & may be used to build the output classification string in the following way: substring& n will be built by finding the n-th occurrence of the character in the classification rule, reading the content of the corresponding column and appending it to substring.

  
Table: MIDAS session for classifying SUSI exposures

Suppose you have created the classification rule WFIB for flagging all exposures obtained in the blue arm of EMMI using the wide-field imaging mode. Now you want to flag these exposures with a character string containing the filter number that has been used.
The corresponding descriptor WFIB of the OST may in natural language be: ``frame exposed in the blue arm using any filter and no grating" The translated selected criterion looks as follows:

:FILTB_ID.EQ."*".AND.:FILTB_TYP.EQ."FILTER*".AND.:GRATB_ID.EQ.""