PDA

View Full Version : Free Linux image stacking software?



JohnOwens
2005-Mar-27, 12:23 AM
Does anyone know of a good solution for doing image stacking under Linux? I've found ImageJ (http://rsb.info.nih.gov/ij/index.html), Java-based, and AstroStack (http://www.astrostack.com/), also Java-based, but AstroStack is mighty slow, running under Java, and has a limit on how much image it'll handle without upgrading to the paid version, and I forget just why, but I found ImageJ to be fairly poorly suited as well. I tried installing the usual Registax (http://registax.astronomy.net/) on my Windows, and running it under Wine, but that gives a trivial error that renders it completely useless (when I try to select the images for the stack, I get a "division by zero" error instead of file selection window). I'm still going to try installing the updated version (when I installed it, it was a beta version) and see if that fixes things, next time I'm booting into Windows.
But, isn't there some decent, free (beer and/or speech), native Linux (or even FreeBSD) software that will do the job? Or a plugin for the Gimp (http://www.gimp.org/), even? I've scoured the Web pretty well, and not found a darned thing.

Robert Andersson
2005-Mar-28, 12:45 AM
Sorry, no, but there must be, right? :D

Have you tried to browse http://freshmeat.net/? Under, eg:
* Scientific/Engineering -> Astronomy (http://freshmeat.net/browse/134/)
* Multimedia -> Graphics (http://freshmeat.net/browse/100/)
* Artistic Software (http://freshmeat.net/browse/901/)

I'm not sure about your exact requirements so I cannot suggest further.

JohnOwens
2005-Mar-28, 01:07 AM
Have you tried to browse http://freshmeat.net/?...
No, I'd tried Sourceforge, and lots of googling, but I hadn't thought of Freshmeat yet. I'll go check that out....

ChibiVader
2005-Mar-28, 02:15 AM
iraf doesn't have an image stacking feature?

JohnOwens
2005-Mar-28, 02:57 AM
iraf doesn't have an image stacking feature?
I think it does? but I was hoping not to have to wade through too many of the other features (also an issue with some of the packages that are intended to control your telescope and take the pictures for you, too), and I think that might have been one that only used FITS images, and I'm hoping for one that will take JPEGs, and hopefully AVIs, too. I did download it and install it, but it's got a bit of a learning curve. (And you almost have to create a new user with the name 'iraf' and its own home directory?? Whose bright idea was that?)

ngc3314
2005-Mar-28, 02:58 PM
iraf doesn't have an image stacking feature?
I think it does? but I was hoping not to have to wade through too many of the other features (also an issue with some of the packages that are intended to control your telescope and take the pictures for you, too), and I think that might have been one that only used FITS images, and I'm hoping for one that will take JPEGs, and hopefully AVIs, too. I did download it and install it, but it's got a bit of a learning curve. (And you almost have to create a new user with the name 'iraf' and its own home directory?? Whose bright idea was that?)

Sure, IRAF is the way to go - if you care about exactly how many sigma to reject above and below the mean, or you need to reject the top 18% at each pixel, or use the median, or care about the exact registration behavior during alignment at the 0.05-pixel level and need to specify the exact interpolation procedure ... you get my drift. (At least IRAF now knows how to import JPEG and similar formats...) Certainly for almost any amateur application, it is much more tedious compared to Mira or Registax under Windows. ("Much" doesn't cover it...) although there are some tricks letting it do the bookkeeping that can speed things along. The holdup is that, as far as I've ever been able to tell while using IRAF since 1985, you have to create a file with sets of registration points for each image, which is done either manually or with a lot of manual help (although there may be a feature now that lets it catalog stars and try to match them).

Gullible Jones
2005-Mar-28, 08:51 PM
Java-based apps are faster with NPTL... I don't think you'd be able to switch to that easily unless you used Gentoo or something though. (And I don't think anyone would run Gentoo on their main desktop box... I tried about 5 times and figured out why. :oops: )

JohnOwens
2006-Aug-02, 03:16 AM
ChibiVader, I'm afraid I'm going to have to hurt you now. I've finally gotten around to (more) seriously tackling this IRAF stuff, along with the associated ds9, x11iraf, xpa, pgplot, etc. etc., and OUCH! :confused: At least I'm fairly sure now that there's got to be some way within it to do stacking, per se; now I'm wondering how manual or automated the registration of the images is. Just hope I don't go off the deep end and start writing my own registration package for it if/when I find out there isn't one already.

ngc3314
2006-Aug-02, 04:03 AM
ChibiVader, I'm afraid I'm going to have to hurt you now. I've finally gotten around to (more) seriously tackling this IRAF stuff, along with the associated ds9, x11iraf, xpa, pgplot, etc. etc., and OUCH! :confused: At least I'm fairly sure now that there's got to be some way within it to do stacking, per se; now I'm wondering how manual or automated the registration of the images is. Just hope I don't go off the deep end and start writing my own registration package for it if/when I find out there isn't one already.

A few hints:

For stacking of registered images, the command imcombine is your friend. You can set upper/lower thresholds to ignore, so you keep information and weight it properly where only some of the images cover an area.

Registration - ahhh, ummmm, I haven't found a way that's nearly as automated as one would like. There are some tricks which will let you refine image shifts at the pixel level if you can get within a few pixels manually.

Other general tricks - most IRAF commands can operate on whole lists of images, or use lists either of images or numerical values, and address them with the syntax @filename where a file or file list would normally be given. These can be created on the fly by other IRAF commands.

You might also check out some of the "contributed" IRAF packages - dimsum (dithered image summation) is designed for stacking large numbers of dithered images such as one needs to do in the near-IR, and might posisbly have a better registration technique than the standard packages.