next up previous contents
Next: Classification of images Up: Classification of Images Previous: Classification of Images

Creation of the Classification Rules

An instruction for grouping flat field exposures could be: Select all files which have a descriptor IDENT matching one of the substrings 'FF', 'SKYFL', or 'FLAT'. With all relevant information being available in an Observation Summary Table, this can be accomplished by supplying a suitable logical expression to the general purpose command SELECT/TABLE. Because observers nomenclatures usually don't follow any standard conventions, the possibility is given to the users to write their own rules using an interface built on top on the Table Editor. This interface enables the user to interactively enter constraints on the existing fields of the OST. Relational operators ( >, <, != or =) may be used as well as logical operators (&, |). Values or ranges of values have to be specified (``*" is the wildcard character, ``~" ignores the case and the ``.." specifies a range of values). Constraints applied to more than one column are ANDed and translated into an expression understandable by the SELECT command. This expression is stored into a descriptor of the OST. The command
CREATE/CRULE susi_ost bias
first creates a temporary table containing two columns labeled respectively :COLUMN and :RULE and then invokes the table editor. The user should then, for instance, formulate a rule for classifying BIAS exposures by entering constraints on columns of the OST (Table 15.4). Finally the command stores the derived selection criterion in the descriptor BIAS of the table susi_ost:
IDENT.eq."*BIAS*".AND.:EXPTIME.LE.1
This rule would lead to all files being classified as bias exposures which contain the string BIAS and the exposure time of which did not exceed one second. The temporary table file is automatically deleted when the descriptor of the OST has been created.
next up previous contents
Next: Classification of images Up: Classification of Images Previous: Classification of Images
Petra Nass
1999-06-15