Author Topic: Star Mask Challenge  (Read 8675 times)

Offline phcjpp

  • Newcomer
  • Posts: 44
    • View Profile
    • Wimbledon Astronomy
Star Mask Challenge
« on: 2013 September 09 11:23:31 »
Hi Guys,

Could anyone assist with a star mask on this Ha image of the Helix ? Really struggling to get all the stars - especially in the Helix itself. There is also a large amount of nebulosity in the background.

https://docs.google.com/file/d/0BwBdynT8ra0scUQ2LXY3QzRCOVk/edit?usp=sharing

67Mb download - about 14hrs of subs in there.

Cheers
Chris

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4105
    • View Profile
Re: Star Mask Challenge
« Reply #1 on: 2013 September 09 11:45:28 »
not at my PI computer right now so i can't try, but one technique is to do a HDR wavelets compression on the image and then try to extract the star mask.

Offline phcjpp

  • Newcomer
  • Posts: 44
    • View Profile
    • Wimbledon Astronomy
Re: Star Mask Challenge
« Reply #2 on: 2013 September 10 02:27:16 »
Hi Guys,

I had some success last night with this problem - I had an hour (12) of 'L' 5min subs lying around that I ran starmask on (following a standard STF->HT stretch). Not sure what people think about this sort of approach ? Starmask (for me) often works better on a stretched image. I think I may have done a small rangemask selection too on the stretched image to help with background rejection.

If I wanted to blur/smooth the starmask further, is there a good tool for that ?

Chris

Offline dgbarar

  • PixInsight Enthusiast
  • **
  • Posts: 91
    • View Profile
Re: Star Mask Challenge
« Reply #3 on: 2013 September 10 06:49:33 »
Hi Chris,

Here is a link that uses the technique pfile suggested for using HDRWT to prepare the image before building a star mask.

Don

http://www.deepskycolors.com/archive/2011/09/08/star-size-reduction-via-Morphological-.html
« Last Edit: 2013 September 10 11:29:43 by dgbarar »
AP 155 EDF
Paramount ME
ST10 XME/CFW 8A

Offline Juan Conejero

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 6582
    • View Profile
    • http://pixinsight.com/
Re: Star Mask Challenge
« Reply #4 on: 2013 September 10 11:22:28 »
Hi Chris,

As others have pointed out, the HDRMT + StarMask combo is a powerful solution for star mask generation. In fact, the next version of the StarMask tool will probably include some dynamic range compression parameters. Here is a quick example with your data.

In general, star masks are made with nonlinear (stretched) images. For this example I have worked with a stretched duplicate of the image. To stretch it, I have just applied automatic screen stretch parameters transferred from STF to HistogramTransformation.

1. Compress the dynamic range aggressively to flatten the image.


2. Some noise reduction will help us to isolate stars from background noise with the StarMask tool. In this case I have applied a simple median filter.


3. This is the result of StarMask with default parameters except noise threshold = 0.2.


4. The mask can be further tweaked to improve star protection. For example, erosion and dilation filters can be applied to shrink and inflate mask structures, respectively. In this case I have just clipped the histogram at the shadows a little, to remove some unwanted structures at the bottom (for some reason (removed gradients?), the image seems to have more noise at the bottom side).


5. The result. Here you can see the mask enabled for the original image (with AutoStretch STF applied).


Let me know if this is the kind of result you were looking for.
Juan Conejero
PixInsight Development Team
http://pixinsight.com/

Offline phcjpp

  • Newcomer
  • Posts: 44
    • View Profile
    • Wimbledon Astronomy
Re: Star Mask Challenge
« Reply #5 on: 2013 September 10 11:46:23 »
Juan - thanks! This is the best star mask tutorial I have seen to date (I have looked at a few) - I imagine your approach can be followed for most images. I will try it myself this evening.

As far as noise / structures etc. in this particular image .....

I don't think this image had had anything done to it before I uploaded it. As stated in the original post its around 14hrs of pre-sorted high quality subs from SSO in Australia. I would be a little surprised if there were any serious gradients of note in the image given its an exceptionally dark site (you get told off by the main observatory for having uncovered LED's on equipment!). There is certainly nebulosity over the full frame though.

Again, thanks so much for taking the time to do this - it really is fantastic.

Chris

Offline Juan Conejero

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 6582
    • View Profile
    • http://pixinsight.com/
Re: Star Mask Challenge
« Reply #6 on: 2013 September 10 12:36:16 »
Hi Chris,

Glad to know it's useful for you. Please ignore my comment about gradients, which has obviously been a digression on my part.

We are working on new versions of some tools these days, including deconvolution tools. So I have been using your image to make some tests, and thought you'd be interested in some results. This is a quick test with DynamicPSF and Deconvolution.

First I have generated a model of the PSF of the image with the DynamicPSF tool. I have selected about 20 well shaped, unsaturated stars. To improve mask protection, I have applied two instances of a dilation filter with a circular 5x5 kernel (MorphologicalTransformation) to the star mask image from my previous post. Finally, this is the result of 40 iterations of regularized Richardson-Lucy deconvolution:

Before deconvolution:


After deconvolution:


You can see the PSF model image at the top right corner of the screenshots above. I have used both global and local deringing (with the star_mask image as a local deringing support), and have also adjusted regularization parameters to avoid noise intensification. Please note that this is just a quick and dirty test, so the result is not "serious". In particular, a mask should be used to protect low SNR areas.

Anyway, the improvement in visibility of small-scale structures and the star size reduction are quite nice IMO. This is a before/after comparison as an animated GIF image:


Juan Conejero
PixInsight Development Team
http://pixinsight.com/

Offline phcjpp

  • Newcomer
  • Posts: 44
    • View Profile
    • Wimbledon Astronomy
Re: Star Mask Challenge
« Reply #7 on: 2013 September 10 12:41:49 »
Thanks again - this is exactly my job for tonight. I messed around for a couple of hours last night to check the quality of the data (attached). Now I will get a little more serious and take my time. Was about to start deconvolution when I noticed some bad pixel rows in my NII - time to put the repair script to work. Something tells me the bad pixel map in Maxim went wrong!

Attached is from last nights session.

Chris

Offline marekc

  • PixInsight Addict
  • ***
  • Posts: 177
    • View Profile
Re: Star Mask Challenge
« Reply #8 on: 2013 September 12 10:25:14 »
Juan, thank you very much for posting your workflow for StarMask-ing this image. I always find that sort of thing very helpful.

I downloaded Chris's Ha image, and I'm trying to follow along with your processing steps for star masking.

I'm having trouble duplicating your result at the HDRMT stage, though. When I apply HDRMT with the same parameters you showed in your example, it doesn't compress the dynamic range in the nebula much at all. It just darkens the background a bit, and creates some dark ringing around the brighter stars. But the nebula remains very bright.

I started the workflow just as you did: I took Chris's linear image, and stretched it with Histogram Transformation. The HT parameters came from an AutoSTF of Chris's image, as in your example.

After that, I set the parameters in HDRMT just as you did: Number of layers = 4, Iterations = 3, Overdrive = 0, Inverted and Median transform UNchecked, Scaling function = B3 Spline, and the three remaining checkboxes UNchecked.

But these settings didn't flatten the nebulosity much, if at all, when I applied HDRMT to the non-linear image.

Had you perhaps made some settings in Midtones Balance that weren't shown in your screenshot? Or maybe I've missed something in the way I'm applying HDRMT.

Thanks,

- Marek

Offline phcjpp

  • Newcomer
  • Posts: 44
    • View Profile
    • Wimbledon Astronomy
Re: Star Mask Challenge
« Reply #9 on: 2014 January 27 06:21:15 »
Hey Guys,

Just thought I would add a couple of links to this thread - processing in links below was not done in PI (except the pre-processing which was done by me by hand in PI) but interesting all the same I think.

http://apod.nasa.gov/apod/ap140110.html
http://astrodonimaging.com/gallery/display.cfm?imgID=265

If people are interested and PI wanted to host I can supply the HA, NII and OIII subs used for this. I would be especially interested if someone wanted to attempt to make a tutorial out of it as I think the data is really very good.

Cheers
Chris

Offline RickS

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1295
    • View Profile
Re: Star Mask Challenge
« Reply #10 on: 2014 January 27 17:07:20 »
Hi Chris,

I'd be interested in having a play with the data.  If you don't mind I'd even like to try combining it with the 40+ hours of Ha and OIII I have collected on the Helix, albeit with a significantly smaller scope (my image and processing notes are here: http://pixinsight.com/forum/index.php?topic=6486.0)

Cheers,
Rick.

Offline phcjpp

  • Newcomer
  • Posts: 44
    • View Profile
    • Wimbledon Astronomy
Re: Star Mask Challenge
« Reply #11 on: 2014 January 28 09:58:53 »
Rick - request granted.

http://pixinsight.com/forum/index.php?topic=6676.0

Cheers
Chris

Offline RickS

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1295
    • View Profile
Re: Star Mask Challenge
« Reply #12 on: 2014 January 28 17:18:27 »
Thanks very much, Chris.  I'm traveling for the next couple of weeks but will have a play with the data when I return.

Cheers,
Rick.

Offline astroedo

  • PixInsight Addict
  • ***
  • Posts: 152
  • Io ne ho viste... cose che voi umani...
    • View Profile
    • L'arciere celeste
Re: Star Mask Challenge
« Reply #13 on: 2014 January 28 22:09:10 »
Hi Chris
I would like to give it a try, not to prepare a tutorial ( I'm not so skilled) but to learn something since I'm about to buy a narrow band set but I have no experience in such a processing.
I will keep you informed about my progresses.

Can I use your data?

Thanks
Edoardo

Offline phcjpp

  • Newcomer
  • Posts: 44
    • View Profile
    • Wimbledon Astronomy
Re: Star Mask Challenge
« Reply #14 on: 2014 January 29 00:07:31 »
This thread is closed - please follow this link for the data.

http://pixinsight.com/forum/index.php?topic=6676.0

Chris