Replaces all occurrences of magic value pixels in an NDF array with a new
Replaces all occurrences of magic value pixels in an NDF array with a new value
If a constant value is used to replace magic values (which will be the case if
Parameter SIGMA is given the value zero), then the same replacement value is used for
both the data and variance arrays when COMP=
"All". If the variance is being processed,
the replacement value is constrained to be non-negative.
Magic values are replaced by random values if the Parameter SIGMA is given a non-zero value. If both DATA and VARIANCE components are being processed, then the random values are only stored in the DATA component; a constant value equal to SIGMA squared is used to replace all magic values in the VARIANCE component. If only a single component is being processed (whether it be DATA, VARIANCE, or Error), then the random values are used to replace the magic values. If random values are generated which will not fit into the allowed numeric range of the output NDF, then they are discarded and new random values are obtained instead. This continues until a usable value is obtained. This could introduce some statistical bias if many such re-tries are performed. For this reason SIGMA is restricted so that there are at least 4 standard deviations between the mean (given by REPVAL) and the nearest limit. NOMAGIC notifies of any re-tries that are required.
The last of the options forces substitution of bad pixels in both the data and variance
arrays. This parameter is ignored if the data array is the only array component within
!) propagates the title from the input NDF to the output NDF.
If the NDF arrays have no bad pixels the application will abort.
Use GLITCH if a neighbourhood context is required to remove the bad values.
This routine correctly processes the AXIS, DATA, QUALITY, VARIANCE, LABEL, TITLE, UNITS, WCS, and HISTORY components of an NDF data structure and propagates all extensions.
Processing of bad pixels and automatic quality masking are supported.
All non-complex numeric data types can be handled.
Any number of NDF dimensions is supported.