imcoadd: workaround to include all pixels?

Adam wrote on Feb 06, 2013

I'm looking for a workaround to the issue in imcoadd where the images are *cropped* rather than expanded to fill the full frame:
http://www.gemini.edu/sciops/data/IRAFdoc/imcoadd.html (see item 2 under Planned upgrades to IMCOADD).

If I have a good geometric fit for my individual subframes, can I use those with some other task to create a complete co-add with the edges expanded?

I'm working on data with 4 subframes that were dithered, and using imcoadd I end up with the chip gaps appearing because at least one of the 5 images did not cover that area.

Thanks,
Adam

---

James Turner wrote on Feb 06, 2013

I don't do much imaging myself, but I was wondering if you might use geotran directly to perform any necessary rectification on each individual image (since you already have the solution for that) -- or start with the intermediate images from imcoadd -- and then co-add them yourself with imcombine? It looks at a glance like that's what imcoadd does internally anyway; it's a CL script so you can see what core IRAF tasks it calls with which parameters, though unfortunately it is 2000 lines long... I'm having trouble spotting quickly where the cropping occurs, since imcombine itself is called with outlimits="" and it looks like the full pixel range is also specified when calling geotran. This would need a bit more investigation. If you can't make any headway with that, you could send your question to the Gemini helpdesk and one of my colleagues who has studied imcoadd more may have better insight.

I suspect the "to do" list you mention is rather old and unless Inger has a renewed interest in imcoadd for some reason, it's more likely that we'll provide greater flexibility in future via our pipeline processing in Python.

Cheers,

James.

---

Adam wrote on Feb 06, 2013

Thanks, that's effectively what I ended up doing, though I ended up using imexam + ccxymatch + ccmap to derive the geometric transform.

As you said, python scripts are the way to go.... code here:
https://github.com/keflavich/irafscripts

---

James Turner wrote on Feb 06, 2013

Thanks for posting that. Just FYI, we're also setting up a forum at Gemini where users will be able to post (or link to) their own scripts for reducing our data.

I'm told it's the geotran step in imcoadd that's doing the cropping and that geotran can't easily be made to do otherwise.

Cheers,

James.

---

Last post on Feb 06, 2013