How to use Dark Field Subtraction to remove fixed-pattern noise

What is noise in an image? Noise is a variation in brightness or colour in an image which is not really present in the object being imaged but is an artefact of the camera sensor and electronics.

Noise appears as small coloured dots appearing in a colour image, or grey dots appearing in a monochrome image. This is normal. CMOS and CCD sensors are affected by Fixed Pattern Noise (FPN) in particular, however there’s also “random” noise which we’ll discuss at a later date. FPN is where certain pixels in a sensor give a consistently different intensity reading compared to their neighbours. The position of the pixels doesn’t change much, but the intensity of the pixels does, making them more obvious. This is especially true, when:

1) the exposure time is lengthened
2) a higher gain setting is used (which serves to amplify pixel intensity) and
3) the warmer the temperature of the sensor (sensors tend to warm up over time)

Altaircapture uses a method which can loosely be called “Dark Field Subtraction” to cancel out the noise in a single exposure image, or even a real-time video capture. A sample set or “stack” of images is combined into a single reference image, which is held in memory or saved on your hard drive to act as a reference. This reference is applied to the image coming from your camera and after and algorithm is run, the FPN in the image is “magically” suppressed. The noise really does disappear as if by magic in front of your eyes!

Because noise is determined by the exposure time, as well as the temperature and the gain settings of your camera, you naturally need to capture your stack of darkframes using the same settings you would when imaging so everything cancels out as neatly as possible.

So for best results, take your darkframes when your camera is set up and ready, before the imaging session. First decide on what settings you’re going to use, and then when you’re happy with the preview, it’s time to start that darkframe stack. You will need to place the lens cap firmly on the camera or telescope, so there’s no light leakage. We want to only sample the noise the camera sensor produces. If you let any light in, it will become part of the darkframe and it will be subtracted from the final image output giving some pretty weird but not useful effects!

You need about 5-10 frames to get a decent darkframe. It might help to try 5 frames for short exposures of say up to 25 secs and perhaps 10 frames beyond that.

You can export and save that darkframe file as a window’s bitmap, and then load it again and re-apply it at any time. If you want to apply it, while the camera is running, just click the “Apply” tickbox after the dark-frame stacking countdown is complete.

Remember if you hit the apply tickbox half way through a 30 second exposure, it will take another 30 seconds after the current exposure is complete for the reduced-noise frame to appear on your monitor. Be patient, don’t think it isn’t working! For a daytime test just leave the lens cap on with exposures of 15 seconds at room temperature, and click apply, then un-click it to see the results.

The realtime darkframe feature allows you to push modern CMOS sensors into new territory. Experimentation is the way to go, and you don’t need to be connected to a telescope to try this out!

Here’s some images to show you how it works with the GPCAM Mono, lens cap on, running video at 20 secs exposure with full gain of 1000x. You can see the “Dark Field Subtraction” menu opened on the left menu stack. The “Capture” button was clicked with the default 10x frames. After 200 seconds the camera complete the capture, and generated the reference darkframe (which can be exported and saved for later use). Here are two screenshots of a 20 second frame with and without the “Enable” checkbox ticked.

Without AltairCapture Dark Field Subtraction you can see small grey dots in the preview pane which are fixed pattern noise:

Darkfield Subtraction & Fixed Pattern Noise
Fixed Pattern Noise visible with CMOS sensor

With AltairCapture Dark Field Subtraction enabled – and look – no more dots! Yes, we know this is painfully obvious – perhaps just as painful as those cleaning product adverts where the superhero waves his magic sponge, and the grime disappears – but it really does look like this – try it!

Darkfield Subtraction Applied & Fixed Pattern Noise removed!
Darkfield Subtraction Applied and Fixed Pattern Noise removed!