DM-8982: Incorrect binning in overscan spline interpolation #25
Conversation
The binning was incorrect in the presence of masked rows, which could cause the ordinate to be non-monotonic increasing, resulting in an exception. The problem was that the masking wasn't being applied in the binning, but it was in the scaling (the 'numBins' vector).
This helps us identify what was clipped.
PaulPrice
force-pushed
the
tickets/DM-8982
branch
from
da9fcc3
to
4e287d2
Compare
| @@ -345,6 +345,8 @@ def overscanCorrection(ampMaskedImage, overscanImage, fitType='MEDIAN', order=1, | |||
| diff = numpy.abs(biasArray - medianBiasArr[:, numpy.newaxis]) | |||
| biasMaskedArr = numpy.ma.masked_where(diff > collapseRej*stdevBiasArr[:, numpy.newaxis], biasArray) | |||
| collapsed = numpy.mean(biasMaskedArr, axis=1) | |||
| if collapsed.mask.sum() > 0: | |||
| collapsed.data[collapsed.mask] = numpy.mean(biasArray.data[collapsed.mask], axis=1) | |||
There was a problem hiding this comment.
I realize you are adding the original values back into the masked pixels in order to plot them below, but might this be a gotcha for anyone later using collapsed.data and expecting the non-modified values? I'm not familiar with the object here, so am not sure what the masked pixels would've been assigned in collapsed.data, so forgive my naïveté if this is a non-issue!
There was a problem hiding this comment.
Without this change, the masked pixels have a value of zero and are therefore useless.
There was a problem hiding this comment.
That's sort of my point, i.e. someone expecting a value of zero in collapsed.data for masked pixels (I don't imagine a use case here...but you never know!)
There was a problem hiding this comment.
I don't know why someone would want a zero value for masked pixels, except for my particular use of this below! I use collapsed.data*~collapsedMask, which provides exactly that (collapsedMask is True for bad pixels). So the possibility is still there just in case.
No description provided.