Introduction to CCD Imaging

by Arthur Babcock and Robin Casady

A. Introduction by Arthur Babcock

Our purpose here is to provide an introduction to amateur CCD imaging. This presentation is not intended to be either a) a theoretical discussion of the workings of CCD cameras; or b) a "how-to" guide to CCD work. This overview is mainly intended for amateurs who have read and heard about CCDs, but who haven’t had any direct experience with them.

B. Brief History

CCD cameras are electronic imaging devices that professional astronomers have been using for about the last 20 years or so (the first astronomical image of Uranus was made in 1975). They have been available to amateurs for almost 10 years (the first Santa Barbara Instrument Group advertisement in Sky and Telescope appeared in December, 1989). They are related to video cameras and medical imagers, with the big difference being that CCDs used for astronomy have to be cooled to perhaps 25 deg Celsius below ambient temperature in order to integrate, i.e., to make exposures longer than a second or so.

I [i.e., Arthur] bought one about 6 months ago. It’s the Meade 216XT, which uses the Texas Instruments TC-255 chip, the same chip as in the SBIG ST-5 and the Celestron Pixcel 255 (which is in fact a reissue of the ST-5). The chip measures 324 x 240 "pixels," the little rectangles that make up the chip.

On the camera body you can see the cooling fins and sockets for connections to the computer. Nearly all CCD cameras for amateurs are meant to be used in conjunction with a personal computer.

The first thing to point out about the chip is its small size. (3.2mm x 2.4mm). This is one of the greatest challenges facing the amateur: inexpensive CCD chips are tiny compared to 35mm film. At first, one of the hardest things to do is to get the object on the chip.

C. OK, so why bother?

If it’s so tiny, and you have to lug a computer outside to use it, why would anyone want one of these things? Here are the advantages:

1. Sensitivity. These things are highly sensitive compared to film. In the CCD world, people talk about "quantum efficiency:" if every single photon that falls on the chip is detected, then the chip has 100% efficiency. CCD chips available to amateurs might have, say, 30% to 50% quantum efficiency. The quantum efficiency of film, on the other hand, is only 3% to 5%! What does this mean? With CCDs, shorter exposures are required to capture fainter objects. Bottom line: An amateur today, working in moderately light-polluted skies, can image fainter stars than the photographic limit of the 200-inch telescope at Palomar!

2. Digital images lend themselves especially well to image processing. We’ve all seen the wonders that NASA does with image enhancement, like fixing the flawed HST images. For that matter, lots of people are out there with Adobe Photoshop, putting Grandma’s face on Grandpa’s body and stuff like that; with CCDs you can do all that with astro images.

3. Linear response. This is probably more of a scientific advantage than one for amateurs, but, basically, in CCDs, a star that is twice as bright produces a signal that is twice as strong. Film is very non-linear: you have to know the response curve of the particular emulsion you’re using, and it’s a whole big deal.

D. Disadvantages?

1. Small size. If you stick a Meade 216 on your 8" f/10 Celestron SCT, your field of view is 5.5 x 4 arc-minutes (for comparison, the Moon as seen from Earth is about 30 arc-minutes in diameter); at this rate, most of the Messier objects won’t fit in your images. Small size also makes it hard to locate your target.

2. Focus is very critical in CCD work; if the focus is way off, there won’t be an image at all. And you can’t look through the telescope to focus while the camera is mounted unless you purchase an expensive flip mirror adapter with a parfocal eyepiece.

3. CCD pictures aren’t pretty. If you want glorious color images of the North American Nebula, don’t use a CCD; use film. (Incidentally, CCDs are inherently black-and-white devices. The usual means of getting color images is the so-called tri-color method, where you make three separate images of the object, through red, green and blue filters, then combine them to produce a color image.) One shot color CCD cameras are just starting to appear, and may become more common in the future. However, they do not have the same resolution as a black-and-white camera.

E. Images

I thought I’d start with the image that convinced me that CCD imaging was something I might get interested in someday. This is from Richard Berry’s book Introduction to Astronomical Imaging, which comes with image processing software and a few sample images. I bought this a few years ago just as a cheap way of finding out about some of this stuff.

One of the images Berry included was of NGC 891. a fairly faint edge-on galaxy with a dust lane running down the middle. I loaded the image and saw something like this [n.b.: the images used here are different from those in Berry's book]:

What galaxy?, I thought, There's no galaxy here! Then, I "scaled" the image, and the galaxy popped out at me:

Aha! There's a galaxy there after all! What this shows is that very small amounts of light falling on a CCD chip are in fact recorded, even if they are so faint, so close to the background level that we cannot discern them without processing. But by "stretching" the gray scale, we can "pop out" and make visible any part of the image that is not absolutely the same as the background. With film, once you get below a certain level, there is very little you can do to "pop out" a hidden image. Images too faint to register on film just aren't there at all.

Images taken with the 216XT

1. The first-quarter Moon taken through a 60mm refractor with yellow filter, 1/100 sec. The first thing to try imaging is the Moon, because it’s easy to find. This was done with a stationary tripod, no equatorial mount. This simplifies things even further.

2. M41 is an open cluster about 4 deg S of Sirius. This is actually the Northwest quadrant—the whole cluster wouldn’t fit on the chip. I’ve reversed black and white because it’s easier for the eye to detect faint levels of black than faint levels of white, and I amused myself by looking up the magnitudes of some of the stars, which I then superimposed on the image. The faintest magnitude that I labeled is 12.6; some of the other stars detectable on the image are no doubt fainter than that. Now here’s the impressive part, that shows how sensitive these CCD chips are: this is 1) only a 25-second exposure, 2) through a tiny 60mm refractor, 3) through thin high cloud, 4) on a night of full Moon!

1. This is M104, the Sombrero galaxy, through a 6" f/6 Newtonian. The exposure is 90 seconds. The faint star in the upper right-hand corner is of magnitude 15.4. This is not too bad an image, but let’s look at the next one. (This is the only special technique that I’m going to illustrate).

2. This image of M104 is three 90-second images stacked on top of each other. You can do this very easily in the software. It has the effect of improving the signal-to-noise ratio: note the smoother background, and note also the new faint stars that are visible below the nucleus of the galaxy: they are not visible on the component images.

Why would you stack images rather than just take a longer exposure? Two reasons: 1) guiding, say, ten 1-minute shots is easier than one 10-minute, and 2) long exposures of bright objects can saturate ("overexpose") the pixels, leading to "blooming" (streaks of pure white emanating from star images). Adding shorter exposures together avoids blooming, but collects the weak signals from faint objects.

3. This image of the globular cluster M3 is another example of image stacking (2 images of 60 seconds each, in this case.)

4. This last image is not a very good one; it is only 30 seconds, and the focus is not perfect (I was working fast because clouds were approaching) but it does shows the supernova in M96 (indicated by the tick marks).

F. Conclusion

In an Internet discussion group devoted to CCD imaging, I read a post from a fellow who said that he had been following the discussion for a while to see if he wanted to get into CCD work. His conclusion was that he did NOT want to get involved in a process that seemed fraught with difficulties and that generated such highly technical discussions. Several other members of the group hastened to reply to him that it wasn't nearly as bad as it seemed, that it is in the nature of Internet discussions to become convoluted, etc. I agree with this latter view: if you expect "point and shoot" ease and convenience, CCD imaging is not for you (arguably, astrophotography isn't for you, either). On the other hand, a graduate degree in physics or electrical engineering isn't necessary. The most important prerequisite in my opinion is the patience to acquire a new technology. As in many endeavors, you may expect to get out of it what you put in.

G. Image Processing in Photoshop

My (Robin Casady) contribution to this article has to do with enhancing images after they are taken. My experience with CCD imaging is very limited. However, I've been using Photoshop for a number of years. The techniques outlined here are very basic and simple to perform.

First we start with a raw image of Gassendi Crater (lower left) and Mersenius Crater. This is how the image looked when it was downloaded from the camera.

The gray levels have been adjusted in Photoshop. Notice that the black is now truly black, and a little more detail can be seen in the mountains near the bottom of the image. The change is subtle, but it makes the image a little richer. The original has a somewhat washed-out look by comparison.

To give the image a little more "snap," a Photoshop filter called "Unsharp Mask" was used. It gives the edges more contrast, and makes the image appear sharper. The amount of "sharpness" added to an image can be adjusted. How much to use is an aesthetic decision. When too much is added, the image may look harsh and some visual artifacts may show up as a granularity in a smooth area. This image shows a little of that granularity around Gassendi.

Making a Mosaic of Several Images

To cover a larger are of the Moon, a mosaic is made of several images. Raw CCD images are stacked in Photoshop as layers. Each image is a layer. Notice the exposure differences with the different images.

The images have been aligned with each other by setting the top image to show as 50% transparent, turning off all other layers except the second layer, and aligning that layer with the top layer. When aligned, the second layer is turned off, the third layer is turned on and aligned with the top layer. The procedure is continued until all layers are aligned.

Next, each layer is evaluated to see if it is worth keeping. There are three things to look for:
1. Does it expand the area of the image.
2. Is it sharper than other layers that cover the same area.
3. Is the exposure within bounds that can be adjusted to match the other images.

Images that don't meet at least some of the above criteria are deleted from the Photoshop document. The remaining layers are ordered so the sharpest images are on top. That way the best images will cover the most area of the final image.

Now the individual layers are processed as illustrated above. The gray levels are adjusted so they all match. An Unsharp Mask filter is used to make them look sharper. And the file is saved as a flat (all layers become one) image.

These images were taken with with Arthur's 216XT camera (and his help) through my Takahashi TSC 225 (9" SCT). I've also done a few images with a 208XT that I purchased for autoguiding. More images can be seen on my web site.