Mastering the Dark (Frames)
A recent NightTimes article by Jack Kramer discussed a variety of activities in which an astronomer can engage on cloudy nights. For those of you who are interested in digital imaging, here is another: create a library of dark frames. (This article assumes you have a PC and basic image processing software.)
Background: All digital imaging devices have the same basic components: a charge-coupled device (CCD) chip and the electronics to read out its data. The CCD is a rectangular array of "photon wells" that work by accumulating electrical charge as photons strike them. The amount of charge for each well is read out ("drained") by the electronics and converted into a digital number. These numbers become the data used for the pixels in the re-sulting image.
Noise: Something else all CCD-based imaging devices have in common is noise. This is often the culprit when a digital image is grainy. There are many sources of noise: the analog-to-digital conversion being performed, var-iations in the sensitivity of the individual photon wells, infrared photons from the heat of the CCD electronics, and cosmic rays. You want to remove all their effects from your images. How? Take a picture of them. Actually, lots of pictures.
Dark Frames: A Dark Frame is a picture of the noise. You keep the shutter closed (or covered) and take an image for the same length of time as the actual light-gathering image of your object. If you have a temperature-controlled camera, then the Dark Frame should be at the same temperature as well. Don't just take one Dark Frame at the combination of temperature and time, take 15-20 and save them on your computer. Given all the opportunities we have for this with our Chicago weather, you will soon have hundreds of Dark Frames. I have a library of Dark Frames, mostly all twenty minutes long, at temperatures of -10° C, -15° C, and -20° C. Taking these frames on cloudy nights helps you make better use of those precious minutes of clear sky. You later simply subtract the Dark Frame from the Data (light) Frame.
Mastering the Dark: My very first CCD images were awful. When I started using Dark Frames, my images became a little better, but were still grainy. This is because subtracting a single Dark Frame from the data frame can actually worsen the signal-to-noise ratio of the final image. The way to improve the S/N ratio is to com-bine the many Dark Frames into a single Master Dark Frame. The simplest way, of course, is just to average them. This is fine until you discover that one or more frames have a hot spot caused by a Cosmic Ray. Averaging will not remove hot spots. Instead, perform a "median combine" to create the master frames. For each pixel, the median pixel value across all the frames is selected. The excessive value caused by the Cosmic Ray will be excluded as a result.
Some tips: If your camera has no temperature control, then build the dark library using ambient (air) temperatures. Record the current temperature when taking the image and use the closest temperature match when processing the image. (You would actually create something called a Bias Frame, but that is beyond the scope of this article.) Another option is to take a Dark Frame immediately before and after the light fame and then average them later, but that can waste a lot of observing time.
Is it possible to create good images without the use of Dark Frames? Of course. The brighter the object, the more the noise will be hidden. This is why digicam images of the Sun, Moon, and planets work very well on their own. As you begin to image faint extended objects you will see the effects of noise and will want to master the dark.
Michael Purcell is the LCAS Webmaster. See his images at http://purcellmf.home.comcast.net/astro/Published in the March 2005 issue of the NightTimes