Do I need to crop before integration ?

[Julien]

Hi,

Do I need to crop (to remove black border) my aligned images before integration ? I mean, does this have any influence in the integration computations ?

Regards

Julien

[pfile]

no - i think ImageIntegration completely ignores black pixels when computing image statistics for integration.

rob

[Niall Saunders]

Hi Julien,

I have always considered Crop to be the first post-processing step after pre-processing. I wouldn't waste the time needed to crop individual frames - after all, if you have aligned these, then ctop them you lose the alignment anyway, and if you crop before aligning and integrating, you will syill need to crop again afterwards.

Hope this helps.

[Julien]

no - i think ImageIntegration completely ignores black pixels when computing image statistics for integration.

rob

That would be nice to have confirmation of that.

[Julien]

if you have aligned these, then ctop them you lose the alignment anyway

No, the alignment is not lost if I crop them with the same coordinates. On some images I will crop more than the black border, but at the end I will have no black border at all on any image

[pfile]

i think for sure with respect to rejection, the way II is configured by default (like when resetting the process) 0 valued pixels are rejected in the outlier algorithms. i have to read the code to find out if the normalization process ignores 0-valued pixels though. the code is on github (search pcl imageintegration).

you are right that alignment is not lost if you apply the same crop to all registered images.

in all my years of doing AP i've never heard of someone pre-cropping their subs to eliminate registration artifacts, so i tend to think it's unnecessary, but just because everyone does something one way does not necessarily make it right...

rob

[Julien]

i think for sure with respect to rejection, the way II is configured by default (like when resetting the process) 0 valued pixels are rejected in the outlier algorithms.

Yes, your right. I just done a test and I can clearly see a white border on the low rejection image.

[chris.bailey]

Rather than pre-crop you can set a region of interest in II. One of my CCD's has distinct edge artefacts with narrowband filters (so called wet edges) so I tend to do II with a region of interest to remove these. Also useful when you are combining data from multiple nights or different camera setups which don't align properly. You can define the ROI from a preview so its quick and easy to see if it makes a difference. I tend to do a preliminary II run with no rejection and define the ROI on based on that.