Stargazing Image Program

Deadly · Mar 4, 2007

I was wondering if anyone knew of a program that would do the following:

* Allows you to input multiple photographs.
* Rotates each of the images to align bright spots (compensates for
Earths rotation and time laps between photos)
* Examines each pixel statistically and determines the most probable
true color to compensate for digital static.

Basically I want to take many photographs using a digital camera and I want the program to use statistics to remove the digital static and improve the final image.

astrostu · Mar 4, 2007

Deadly said:
I was wondering if anyone knew of a program that would do the following:

* Allows you to input multiple photographs.
* Rotates each of the images to align bright spots (compensates for
Earths rotation and time laps between photos)
* Examines each pixel statistically and determines the most probable
true color to compensate for digital static.

Basically I want to take many photographs using a digital camera and I want the program to use statistics to remove the digital static and improve the final image.

Based upon what you seem to want for an end result, I think what you want (assuming you have a PC) is Registax. Since I'm on a Mac, I can't tell you for sure, but I know that it's what many amateur astronomer-astrophotographers use.

I don't know if it compensates for rotation, but I'm sure it compensates for translation.

I doubt it does your third point, but what it does do is take the average of each pixel location (once it's aligned the images) across all the images in order to get the most likely true color. I think it would be close to impossible to do your third point with just a single image unless you make a lot of assumptions that I wouldn't be comfortable with a program making. Just taking the average (or median) is what's usually done.

Deadly · Mar 20, 2007

Thats not what I was looking for. On a side note I checked out your web site, and the first image I saw looked very familiar. I went back and looked at a photo I took a year ago. I think it might be the same binary star.

Focal Length 16mm
F-Number F/4.8
Exposure 15 s
Time/Date 6/24/2006 10:25PM

I don't really know what I'm doing... yet, but this is very cool. If this was the same binary star I guess we could determine the orbital period by comparing the two photos.

You can see the digital static in the background of that photo. At least that is what others told me it was. I guess it could be stray photons diffusing through the air or possibly light pollution. I decided that there was two ways to decide if the speck of light was a star (a real object) or not. One was to allow enough time for the stars to move. Any object that doesn't have a smear is probably static. The other was to compare photos and see if the light spot appears frequently in the same spot in every shot. The only problem is that the earth is moving so you would have to compensate for that. After you did that you could go through and check every single pixel of say n number of photos. Only keeping the most probable "true" color. IDK just an interesting idea.

Deadly · Mar 20, 2007

Maybe not they seem too far apart to be Albireo.

astrostu · Mar 20, 2007

Deadly said:
Maybe not they seem too far apart to be Albireo.

Since there's no scale given, there's no basis for this claim.

Also, I'm pretty sure Albireo is just an optical binary, not physical, so there's no orbital period to discern.

As for the "digital static," I still maintain that the easiest way to get rid of it is to average several images. Since digital noise is generally random noise, then averaging it together will make it flat. For example, roll a 6-sided die 10 times. Then average the values on each roll. Though the individual value of each roll was completely random, the average of multiple rolls should be around 3.5.

If it's really a star, then it will remain visible, just like if you were to roll a 6-sided die that instead of having 1-6 on it had 2-7. The roll values would still be random, but you'd average out a value of 4.5 (a very faint star). A very bright star would be like rolling a 6-sided die that had values 101-106 on it. The noise level would still give you the same fluctuation as before, but since the "possible" values are so much higher than that noise level, you don't really have to roll multiple times to see that it's bright (high number in this example).

The other method you're suggesting sounds much more complicated.

Edit: Wikipedia says I'm wrong and that it's a physical binary. 385 light-years from Earth and 34 arcsec apart means that they're a minimum of 7 light-years apart, so one orbital period is incredibly long and there's no way we could detect it with observations less than a year apart.