Combining different binning data with Pixinsight
Sometimes, with our limited amateur equipments, we are trying to capture the details and the lowest SNR objects in the background. There are some sources, like diffuse nebula, IFN and narrowband data where capturing at 1x1 is no benefit at all. No matter how hard we try, the readout noise is extremely big to capture well this faint objects.
Are we going to get enough data for catching details in 1x1 binning?. The answer is no. This is the moment where binning 2x2 and longer exposures come into play, we can go a lot further in less time, and capture the data in those bad seeing nights.
I faced this problem while imaging a planetary nebula with faint nebulosity in the background. The 3nm Ha filter was not helping at all, and my optimal subexposure time was about 5 hours in bin 2x2. As this is not possible because of meridian flips and mental sanity, I decided to go with 1 hour exposures in bin 2x2.
Well, we are done, I have the nebula captured in 1x1, where stars are tight and some details are better than in the 2x2 capture. I also have the nebula in the same capture time, but in 2x2: bloated stars but much less noise overall. But, how can we get the most of both images?. A difficult task, indeed.
This is my particular solution for the problem:
1) Deconvolution for both images. My trick works with non linear images.
2) Align the images, this will also duplicate the size of the 2x2 binned image.
3) Linear fit them. This is an easy task, because both images should be easily fitted. At least in narrowband with the same filter.
4) Apply THE SAME histogram transform to both images. As they are linear fitted, both should be identical.
5) Create a mask for the 1x1 image, clone it and apply a ATWT with 1, 2, 3, 4 layers disabled. This creates a map of the high SNR zones, and the lower zones, erasing the noise from the equation. The details are also out of the equation. Perfect mask for the job.
6) Apply that mask inverted to the 1x1 image. We only want the lowest SNR parts affected.
7) Now apply a pixel math simple equation:
bin1x1Image + bin2x2Image * MULTIPLIER
Experiment with the multiplier, it will balance the amount of 2x2 data added. The rescale result must be activated.
And here you can see a result, I think is pretty good!.