not a bug LocalNormalization and pedestals may lead to problems

const

Active member
  1. Have a set of aligned files with PEDESTAL tags. It can be the same value across all files or different values, like created with Auto feature in ImageCalibration.
  2. Run LocalNormalization on these files.
  3. Run ImageIntegration with .xnml files and Subtract pedestals enabled — pedestals effectively double subtracted.
Only if all input pedestals are equal and you do not enable subtraction, the result is meaningful. LN is applied correctly and the output file has the same PEDESTAL value assigned.
If input pedestals are different, the output file has no trace of them, but at least the data is valid.
I see two possible ways to make it better:
  • Make LN aware of pedestals and compute corrections assuming that II will later subtract potentially different pedestals. The Subtract pedestals option should be forced by II when Local normalization is used.
  • Grey out the Subtract pedestals option in II when Local normalization is used. The output will logically have the pedestal of the LN reference, which is hard to know at this stage. So better not issue the PEDESTAL tag in the output.
 
Option 'Subtract pedestals' is an exceptional case, only applicable for integration of calibrated light frames when a global pixel rejection algorithm is applied. The behavior of the ImageIntegration process is described in the tooltip of option 'Subtract pedestals':

Subtract PEDESTAL keyword values, if present in input images.

Some data sets require the addition of small positive values (typically about 100 DN) to calibrated light frames in order to prevent truncation of negative pixel sample values after calibration. These small pedestals can be subtracted from source integration pixels to refer all input data to the same zero point consistently.

In general, this option should only be enabled for integration of calibrated light frames with PEDESTAL keywords when a global normalization method (such as additive with scaling) is used for pixel rejection and/or integration output.

This option should always be disabled if local normalization is used, since subtracting pedestals invalidates local normalization functions. Local normalization already compensates for differing additive components among integrated frames.

This option should also be disabled for integration of raw bias and dark frames, since in these cases existing PEDESTAL keywords should be preserved in the generated master bias and dark frames. These pedestals will be subtracted automatically by the ImageCalibration process for calibration of flat and light frames.

Bernd
 
Back
Top