Transcript
Astronomical Imaging
Astronomical Imaging
Signal-to-Noise Ratio
Author: Daniel Duggan
Astronomical Imaging - Faulkes Telescope Project
Astronomical Imaging Signal-to-Noise Ratio What is Signal-to-Noise Ratio? Signal-to-Noise ratio (often abbreviated to S/N or SNR) is a measure of signal strength relative to background noise. In astronomical imaging, the signal is in the form of light from the object being exposed, and background noise is often caused by background radiation, as well as thermal and electrical noise from the CCD (Charged Coupled Device) camera. The signal and noise strengths vary across a whole image, if it didn't, an image would never form. The pixel values would be the same and you would only see a square of white, grey or black depending on the recorded values. Hereon when SNR is mentioned, it is referring to the average SNR of an image. The incoming signal strength of an image comes from the object being imaged, this can be represented by Intensity Strength, Is; and the background noise can be represented by Intensity Noise, In. If Is = In then SNR = 0. In this situation the signal will be almost unreadable because the background noise strength is just as strong. The object in an astronomical image will be almost indistinguishable from the noise.
Astronomical Imaging
To form a high quality image, Is needs to be greater than In so that the resulting SNR is positive. This will results in an object being more clearly defined and visible against the noise. If Is is less than In, then the SNR is negative. In this situation, the noise strength is greater than the signal strength and image quality will be extremely poor. The object may not be visible against the noise at all. Astro-imagers always endeavor to maximise the SNR of their images, and there are many different ways to do this. Internal thermal and electrical noise can be minimised by cooling the CCD to a very low temperature - some systems get close to absolute zero (-273 oC or -459 oF). Another method is to stack multiple exposures of the same object...
Stacking Stacking images is a process by which at least two images of the same object are taken, and then added together. This adds the pixels values in both images together, to make a third image which is twice as strong (i.e. the pixel values are doubled (assuming the same exposure time on both images)). However, as you might expect, if the signal strength is doubled by adding the images together, then the noise strength must also be doubled. If which case the actual image quality is not increased, as the ratio is the same. Page 1 of 2
Astronomical Imaging - Faulkes Telescope Project
Well this is true, but the stacking process works due to a single fundamental concept: Noise is random. When two images of a galaxy, for example, are stacked, the signal from the galaxy is always coming from the same places - the structure of the galaxy itself. So when the images are stacked the pixel values from the galaxy are combined very successfully. However, the background radiation noise falls on the CCD in a random fashion - it does not appear in the same places and intensities all the time in subsequent images. Therefore when the images are stacked, the noise does not add up as quickly as the signal strength, and this allows for the SNR to increase with every stacked image (assuming the images being stacked have a positive SNR to begin with).
Astronomical Imaging
Below is an example of one 60 second exposure of M51 (left), against four 60 second exposures that have been stacked (right).
Hopefully, you will be able to see that the stacking process has increased the SNR so that the signal strength dominates the image, and noise levels are very low. The effect is most notable in the darker regions of the spiral arms.
Page 2 of 2
