What is Binning? “Binning” is when you combine the signal from adjacent pixels in a camera, which reduces the resolution of the image.
For example 2×2 averaging binning combines 4 pixels into one “virtual pixel” which contains the average values of these four pixels. Additive binning does the same but it adds all four values together. 3×3 Binning give you the same but with 9 pixels and so-on.
What is the effect on resolution? Binning reduces the number of pixels in the final image, so the “resolution” decreases (in other words your 2×2 binned image is the same field size but it now has a quarter of the number of pixels making it up compared to an un-binned image. In some cases, as with modern high resolution CMOS cameras, a reduction in resolution isn’t a problem because there are more than enough pixels to account for the lesser resolution. This can for example by useful if your microscope has a high power objective which your camera out-resolves, or you have a telescope with lower resolution than your camera (oversampling).
Why use binning? An important reason for binning is to reduce the noise in an image, or rather to improve the Signal to Noise Ratio (SNR for short). In effect, we are trading spacial resolution for a better signal-to-noise ratio.
What’s the difference between CMOS camera binning and CCD camera binning?
CCD cameras do “hardware” or “analogue” binning where they add the electrical charges on a set of adjacent pixel up on the sensor, and read that group all at once, before analogue-to-digital-conversion of the charges to a digital brightness value. Each time you read a group of pixels you introduce “read noise” so a component of the value you get is random noise. This means that on a CCD, you are only introducing one “hit” of read noise for each frame, which is composed of the signal for additive-binned pixels. Because CCD cameras have a comparatively higher read noise, usually around 7-11 electrons, this can result in a drastic reduction of the noise component per pixel – a quarter.
CMOS cameras cannot perform “hardware” binning in the same way CCD cameras do. All CMOS cameras use software or “digital” binning where the values of a pixel are either added up or averaged to make a larger “virtual” pixel. This is done off-sensor which is why it’s called “digital” binning or “software” binning. This results in a 50% improvement of the ratio of read noise to signal. Because CMOS sensors have an extremely low read noise (often a couple of electrons or less), the end result is almost always significantly better than a CCD.
Is CMOS “digital” binning inferior to CCD “analogue” binning? In a nutshell, no it isn’t inferior, because CMOS cameras have lower read noise to start with than CCD cameras, so although the noise reduction isn’t as dramatic, it still results in a better SNR than a CCD camera can deliver even if binned.
What the difference between averaging and additive binning? Altair GPCAM and Hypercam CMOS cameras use both averaging or additive binning.
Averaging binning: 2×2 Averaged binning results in a “virtual” pixel with the average of the 4 values.
Additive binning: A 2×2 Additive binned virtual pixel has all four values added up – increasing the brightness, because the higher the number the brighter a pixel is.
Do averaging and additive binning both improve signal to noise ratio? Yes! CMOS-style digital binning reduces the noise level with averaging, rather than increasing relative pixel brightness like addition does, but in fact summing the brightness does the same thing, because data from all four pixels is combined. Therefore, both techniques give the same increase in signal-to-noise ratio, which is the ultimate aim of binning.
Is additive binning useful, considering the differences between CMOS and CCD? Some might say additive pixel brightness increase is an important feature of CCD binning. Brightness however is not that important in itself, because it’s really just a number (an increased value per pixel). You can get the same effect with gain, and in any case, brightness can be manipulated in post-processing, during “stretching” of the pixel values in an image for example: Brightness can be increased proportionally until a dim image feature appears.
What if I turn up the gain in averaging binning mode? The fact that CMOS cameras use “averaging” means that it’s often a good idea to turn up the gain more to “brighten” the image when you use averaging binning. Consider this: Since the amount of noise in the image has already been reduced by binning, it will be easier to process the image than if you turn up the gain the same amount without binning. This may be useful for live video astronomy and EAA but for general imaging, be careful not to overexpose and “blow out” the highlights in the image.
So what is the downside of binning? You are sacrificing spatial resolution for an improvement in signal to noise ratio, therefore your image will have less detail. This doesn’t matter if your telescope or microscope is “oversampling” – in other words the camera has such smaller pixels than the optics can resolve.
Should I bin when doing “lucky” imaging? Opinions really vary on this, because it can be argued that stacking thousands of frames does in effect reduce the signal to noise ratio, and that stacking software may work better with full resolution un-binned images. There have been reports of artefacts (stacking processes are complex), but we think the only way to know for sure is to try it…
We recommend reading this post by Robin Glover (SharpCap author), for a more detailed mathematical explanation of CMOS Camera Binning…