[13] The Horsehead nebula in Orion, photographed by Ian Sharp using an 80mm
refractor, with an ATIK 314L cooled CCD camera.
[14] The Beehive Cluster (Messier 44), also known as Praesepe (Manger), lies at the heart
of Cancer.13Noise
Digital imaging has revolutionised astrophotography in many ways but not
everything in the digital garden is rosy. One major issue with digital imaging
systems is noise which manifests itself as stuff in your images which really
should not be there. There are many different causes of noise but the types
break down into one of two categories: random noise, and static noise.
Static noise appears as a fixed and more importantly, predictable addition
to an image. An example of static noise would be a pixel in your camera
sensor that’s permanently in the on state, known as a "hot pixel". The
simplest way to deal with static noise in images is to follow the normal
exposure(s) with a similar length of exposure with the lens cap fitted. This
creates a "dark frame" which records just the static noise in an image.
The dark frame image can then be subtracted from the normal exposure
(known as a "light frame") correcting the static noise issues. It’s important
to take darks before, after or even in-between normal exposures so the
temperature of capture is the same. The camera settings should also be
kept constant.
Random noise is less easy to deal with and unfortunately also affects
dark frames as well as light frames. It has numerous sources, and is
especially affected by heat which causes false signals to be recorded
across an imaging sensor array. One obvious way to reduce heat related
effects is to cool the imaging chip, and this is exactly what happens in
a cooled astronomical CCD camera. However, there will be a random
pattern of very weak signals lurking in the background of the image.
Part of the process of massaging the image into a visible form is to
stretch the data within the image file – basically making the brightest part
of a faint image as bright as it will go while keeping the dark parts close
to black. If there is random noise lurking in the shadows of an image,
stretching the image will increase its visibility too. The way to deal with
random noise is to take lots of exposures and average them together.
This strengthens the appearance of the subject while helping to smooth
out the random background noise.Optical noise occurs because of defects in the optical path leading up to
the camera. This is typically caused by dust and obstructions to the light
path. A commonly encountered problem occurs when the light cone coming
from the optics is clipped by the edge of the eyepiece holder. This produces a
graduated darkening towards the edge of the frame known as vignetting.
The effect of optical noise can be removed from the light frames
by taking a flat field image, also known as a "flat". A flat is taken by
exposing the telescope to an evenly illuminated light source and taking
an underexposed (approximately to half saturation) shot of the view.
This is then divided into the light frame, removing the optical noise in
the process.
The process of applying all of these corrections is called calibration and
is a necessary evil if you want the best results when photographing deep
sky objects. Although it sounds complex, the process is not too difficult
once you get into a routine. The degree to which you’re prepared to go in
terms of calibration is a personal choice and it’s certainly possible to get
impressive results without going through all the stages. However, if you
do apply calibration, the results will typically be much better.
A typical deep sky imaging session consists of the following stages:1) Set up
2) Capture
3) Calibration
4) Assembly
5) Image processing1) The setup stage is self explanatory and covers the process of setting your
telescope up, finding the target, framing the target and focusing.
2) Capture is the process of taking the light frames, dark frames, flats and
other calibration files should you wish to use them. This stage typically
involves taking a number of light frames of the same exposure length.
These are sometimes referred to as sub-frames and the greater the
number taken, the more random noise reduction can be applied. Random
noise reduces as the square root of the number of frames captured.
For example, if you take 4 light frames, averaging them will reduce the
amount of noise by a factor of 2; the square root of 4. Take 9 light frames
and averaging them reduces the random noise by a factor of 3 and so on.
Following the capture of the light frames, the telescope should be
capped and a number of dark frames taken of the same exposure length
as that used for the lights. These will contain their own random noise
so once again, these need to be averaged to produce what’s known as a
"master dark". The more dark frames you average, the more the random
noise they contain will be smoothed out.
The flats are trickier to achieve because in order to take them you need
to be able to point your scope at a uniform light source. A simple way to
achieve this is to wait until morning and cover the end of the scope with
a taught bit of white cloth, for example from a white tee-shirt. Point the
scope at an evenly illuminated patch of cloudless sky and take a photo.
The photo needs to be between a third and a half exposed for the bestAstronomer Book