Combining slices loses FITS header data

[Bill Tschumy]

I have been investigating using AIJ as a replacement for ASTAP in my photometry processing.  I use a relatively small scope for imaging and commonly stack images to increase the SNR.  I then submit the stacked image to AAVSO for analysis in VPhot.

It seems like AIJ wasn't really designed for image stacking like I need.  After aligning my images, I chose "Combine stack slices into single image...".  This seems to lose all the FITS header information.  This info is needed to work with VPhot.  I could copy the header from a sub-image and apply it to the stacked image,, but this isn't really right either.

Am I doing something wrong or is AIJ not really designed for what I need?  I realize it's main use if time series analysis and it works great for that.

Bill

[karenacollins]

Hi Bill,

It's actually the underlying ImageJ that isn't designed to handle fits header updates as operations are done on the images. The code used to combine the images is currently taken directly from ImageJ, so it isn't fits header aware and ends up dropping the header completely. Nevertheless, we are slowing trying to enhance the underlying ImageJ features to make them fits header aware, and this would be a good next candidate to work on. That said, AIJ will create a minimal fits header when the fits file is written to disk.

One reason we haven't done made this feature fits header aware already (beyond not having the resources) is that I'm not sure what header entries should be maintained and/or created, and how to create those from the individua images. Do you have a sense for what minimal set of headers that Vphot needs?

I'll also mention that Dennis Conti recently contributed AIJ code that creates an "AAVSO variable star report" from AIJ Multi-plot Main. I don't know if that would make AIJ directly capable of doing what you need to do in Vphot,  but I'll get the new version pushed to the AIJ Update server in the next couple of days.

In the meantime, we could continue to discuss how what fits headers should be created.

Karen

[Bill Tschumy]

Karen,

Yes, I noticed the 'minimal" header that is created when writing the combined image to disk.  I think it was 6 lines long ;-)

I'm no expert, but I have a sense of what should be preserved.  As mentioned, I am currently using ASTAP which seems to handle this correctly.  I know you've worked with Han (the author) before and he could probably help here.

I can provide you with the header of an unstacked image and the new header after stacking.  By comparing them you could get a sense of what should be preserved.

T_Cas_Light_V_10s_001.txt
!T_Cas_V_stacked.txt

I *might* be able to help with this since I'm the one requesting it.

[Bill Tschumy]

Actually here is a better comparison.  This is the header from ASTAP after a calibration and alignment:

T_Cas_Light_V_10s_001_aligned.txt

Here is the header after the 10 calibrated and aligned images were stacked.

!T_Cas_V_stacked.txt

As you can see, most info is preserved.  It looks like the DATE-OBS was replaced with the average of the DATE-OBS of the subs.  The EXPTIME was replaced with the sum of the individual exposure times (not sure if that is really correct, but not critical).

He also adds:

JD-AVG  = 2.459822664406E+006         / Julian Day of the observation mid-point.
DATE-AVG= '2022-08-31T03:56:45'                                                

Looks like the WCS data is changed slightly.  Not sure if he averaged these in some way or what.  We could ask Han what he did.

[SCO]

Karen,

I noticed the following statment in your recent message:

"I'll also mention that Dennis Conti recently contributed AIJ code that creates an "AAVSO variable star report" from AIJ Multi-plot Main."

I was not aware of Dennis Conti's work mentioned here.  Could you please elaborate on what this new function does, how it can be used, and if there is another thread that discusses this new function?  Thanks!

Ed

[karenacollins]

The headers suggest that the combined version was re-solved by ASTAP to get the new headers. I think it would be reasonable to copy the first aligned WCS solution if it exists (since they are aligned already). The user could elect to re-solve if desired after the combine.

[karenacollins]

Ed, Dennis has a user guide link in the new panel. Best to check that out when I get it released.

[Bill Tschumy]

Karen,

Might be worth asking Han about that.  I’m sure he has giving it a lot of thought.  My first impression is that we should leave it unsolved and let the user re-solve it if need be.  Using the first WCS solution has the risk of not being entirely accurate for the stacked image and the user would have no way of knowing that.  Leaving it unsolved is the safest solution.

As I mentioned, I would be willing to take a crack at implementing this.  However, you might want to have someone more experienced with the code do it.  That’s fine with me as well.  Let me know.

Bill

[karenacollins]

Hi Ed, Dennis' new AAVSO variable star report functionality is now in AIJ v5.1.1 on the update server. You will see the new feature in the Multi-plot Main > File menu. Dennis provides a link to his user guide on the set-up panel.

[SCO]

That is great!  Thanks Karen.

Ed