Cosmic Imaging: Understanding signal-to-noise ratio

Top-notch processing involves accentuating the positive and minimizing the negative.
| Published: August 23, 2010 | Last updated on May 18, 2023
In my initial column last month, I talked about why I prefer charge-coupled device (CCD) cameras over hand-held digital ones. This month, I continue my discussion of CCD imaging by examining something called the “signal-to-noise ratio.”

Astroimagers use the term “signal” to denote the data — the good stuff that we want in our images. Unfortunately, mixed in with the signal is noise, and it’s always present. Noise can come from many sources. Two are your CCD camera and your computer. Luckily, most noise is typically random. This means the more frames you have to average together, the more the random data smooths out.

As it turns out, combining multiple images is highly superior to using just one image. There is a simple formula for this action: The gain in the signal-to-noise ratio is equal to the square root of the number of frames averaged. That means combining four frames will double your signal-to-noise ratio (2 is the square root of 4), and using nine frames will triple your signal-to-noise ratio (3 is the square root of 9).

Combining multiple images is highly superior to using just one image.
Combining images produces a superior signal-to-noise ratio. Compare the single 450-second image on the top to the one on the bottom, which resulted from combining (stacking) eight 450-second images.
Tony Hallas
Combining images produces a superior signal-to-noise ratio. Compare the single 450-second image on the top to the one on the bottom, which resulted from combining (stacking) eight 450-second images.
Tony Hallas
As you go out further and further, you’ll see that to get a fivefold increase, you have to shoot a whopping 25 frames, and a sixfold gain requires 36 frames! Obviously, there’s a diminishing rate of return for the number of frames used. Also, when you use dozens of frames, other noise begins to enter the picture. I typically try to use nine frames for each of my images. I usually will shoot more than nine frames, but I only process the best nine.

Everything I have noted so far can be lumped into the process called “reduction of data.” It’s the step prior to image processing. You have your collected data, and you want to make it as good as it can be before you actually process it into an image.

All CCD cameras need to have the data reduced in some manner before an imager can process it. It is vital to master this step so that your data is the best possible. You don’t want to hear, “Garbage in, garbage out.” Instead, you want well-reduced data that, when taken into Adobe Photoshop®, will come alive with amazing detail.

Dark current (top), which Hallas described in last month’s column, introduces electronic noise into astroimages taken with CCD cameras. Luckily, it’s easy to remove with software. The result (bottom) speaks for itself.
Tony Hallas
Dark current (top), which Hallas described in last month’s column, introduces electronic noise into astroimages taken with CCD cameras. Luckily, it’s easy to remove with software. The result (bottom) speaks for itself.
Tony Hallas
I hope these first two columns have taken some of the mystery out of the CCD camera. You’ll learn more if you are fortunate enough to have an active astronomy club nearby. There always seems to be imagers among the members, and I highly recommend spending some time with them. They will have hands-on suggestions about good imaging equipment — not just cameras, but also telescopes and mounts — as well as firsthand comments about what software they like.

It’s always a good time to talk about astroimaging, but now is especially so. Prices for entry-level CCD cameras have dropped, and you don’t need a high-end camera to produce amazing images.

In my next installment, I will explore how to use Photoshop to stretch the data you collect, and I’ll explain why I use some of the techniques I do.

Read more of Tony Hallas’ Imaging the Cosmos
September 2010: Getting clean images