This pull request ensures that masking works for ellipse fitting correctly.
Previously, if a pixel evaluated for ellipse fitting computed one or more values for a neighboring pixel that was masked (and therefore has a value we distrust), it was still included in the likelihood calculation.
This PR ensures this behaviour no longer occurs and that any such pixel is omitted for all calculations.
This pull request ensures that masking works for ellipse fitting correctly.
Previously, if a pixel evaluated for ellipse fitting computed one or more values for a neighboring pixel that was masked (and therefore has a value we distrust), it was still included in the likelihood calculation.
This PR ensures this behaviour no longer occurs and that any such pixel is omitted for all calculations.