Light Vortex Astronomy
  • Blog
  • Astrophotographs
  • Articles
  • Tutorials
  • Links
  • About Me
  • About e-EyE
 ​Send a Gift via PayPal:

Remote Hosting in Spain:
Picture

 

Improving SNR in M101 Pinwheel Galaxy in Hydrogen-Alpha 3nm

2/5/2016

4 Comments

 
​I have recently been imaging the M101 Pinwheel Galaxy in Hydrogen-Alpha, with a 3nm filter. Since this is going to be a serious image with tons of exposure time, I decided to do 50 exposures of 15 minutes each at 1x1. The result of stacking all these exposures together has a very, very clear effect on the Signal-to-Noise Ratio (SNR) of the overall result. The following is a GIF animation showing the result of stacking exposures from 5 to 50 in steps of 5​ (click the animated GIF to see a larger version):
Picture
The individually-stacked images were pre-processed in exactly the same manner, using Linear Fit Clipping as the rejection algorithm in ImageIntegration in PixInsight. Once that process was ran, DrizzleIntegration was used to produce these individually-stacked images presented above. Each one was cropped identically using DynamicCrop and non-linear stretched with HistogramTransformation (using the same stretch on all the images). No noise reduction measures or further post-processing was carried out. 

Above, we can see how quickly the SNR improves as we approach 20 exposures in the stack. Beyond it however, there are only minor SNR improvements visible, particularly on the very faint spiral arm fine details (appearing above the noise floor). The result of diminishing returns is clearly presented above.

Everything is better with numbers so I thought it would be a good idea to measure the SNR of these images. I did this by using the NoiseEvaluation script in PixInsight and multiplying the resulting values by 65,535 to convert to 16-bit pixel value. I then divided the Mean of the images (found using the Statistics​ process) by their corresponding noise values to find an estimate for SNR. Below is a graph displaying the results:
Picture
Above each data point is the SNR calculated to three decimal places, using the aforementioned method. To the top-right of the graph is the equation of the exponential curve of best fit, which was plotted to R = 0.9999. This equation demonstrates that the SNR of the image is indeed approximately proportional to the square root of the number of exposures used in stacking. 

​Interesting as this may seem, I wonder if the rule holds as closely for more expansive targets (large nebulae that cover most of the image) and for other filters that are more permissive of light (e.g. Luminance​). Definitely something to test!
4 Comments
Peter Nerbun link
4/5/2016 22:50:44

Very interesting Kayron! If you get a moment I'd be interested in knowing what the histogram peak level was for a single one of your linear images. I've read that levels of between a 20% histogram peak and a 40% histogram peak are ideal for separating the faintest object detail in the galaxy (most likely found in the spiral arms of the galaxy) from camera noise.

Reply
Kayron link
19/6/2016 01:45:56

Hi Peter,

Thank you for your comment. Please be aware that I do intend on replying to your latest e-mails! :)

I've opened one of my M101 HA 15 minute exposures in PixInsight, drawn a small preview box over the galaxy and used the "Statistics" process. It tells me my maximum is "5056".

I'm not 100% what the theory is behind that. What I know is that when one is stretching an image to non-linear, the peak of the histogram being on about the 25% mark (towards white) is where you get the best contrast. Is that what you were referring or is what you read related to image capture in itself? I'm interested too, hahaha! :)

Best Regards,
Kayron

Reply
Peter Nerbun link
19/6/2016 07:14:45

Hi Kayron,

I'm always appreciative to hear your replies to my questions so thank you in advance for any additional reply you have the time to write; you've helped me in so many ways to understand the many subtle techniques for processing deep sky images. I just can't thank you enough for all the time you've given to write answers!

As for my original comment in this thread I was referring to the peak of the histogram in linear (unstretched) images. The way I do that is to open the linear image in Pixinsight and then open the Histogram Transformation tool; once I've done that I select the file name for the image I opened and use my mouse cursor to place the vertical line superimposed on the histogram so that it intersects the peak level on the histogram. Then I look at the value for "x" beneath the histogram, as an example I might see x=0.1977(50-50) corresponding to an "x" axis value for the vertical line that intersects the peak; that means that the histogram peak occurs at about 20%.

Reply
Peter Nerbun link
20/6/2016 06:50:50

I re-read the article about the ideal 20% to 40% range for peak histogram levels and found that it wasn't talking about histograms of the raw linear image but rather of the histogram one sees on the back of a DSLR which utilizes a gamma curve to spread the data across the entire frame in a JPEG format; the article says that as long as the left side of the histogram is separated from the Y-axis by a "reasonable" distance the resulting image won't be lacking for detail at the darker levels of the histogram.

Reply



Leave a Reply.

    Archives

    July 2019
    January 2019
    December 2018
    October 2018
    September 2018
    August 2018
    April 2018
    March 2018
    January 2018
    November 2017
    October 2017
    August 2017
    July 2017
    June 2017
    May 2017
    April 2017
    March 2017
    February 2017
    January 2017
    December 2016
    November 2016
    October 2016
    September 2016
    August 2016
    July 2016
    May 2016
    April 2016
    March 2016
    February 2016
    January 2016
    December 2015
    June 2015
    May 2015
    January 2015
    October 2014
    September 2014
    August 2014
    July 2014