Monday, September 2, 2013

Hottie Pixels and Processed Images

I've been lazy about image processing. To get nice astro images, it takes more than aligning and summing multiple images. Why?

CCD images have artifacts.

Here are the problems:

  1. Hot pixels - These are bright dots that appear in the same spot on each image.
  2. Vignetting - This is uneven illumination of the field.
  3. Dust spots - These end up looking like unfocused blobby blobs.
  4. Noise - Read noise is what gives a grainy appearance to an image.

Above: M57 (Ring Nebula). 15 2-second exposures with Opticstar CCD on 14-inch w/ focal reducer. Fully processed.

What to do? Let's start with the noise. Electronic noise can be subtracted out with a "bias frame". A true bias frame is an image of zero exposure time where the CCD is read out without having been exposed to any light (scope is covered). This allows you to isolate the effect of read noise. The shortest exposure time I can get with camera control software I have (Nebulosity) is 10 ms (0.01 seconds), so this is what I use for my bias frames. It's best to take many bias frames and average them to make a master bias frame.

Above: M31 (Andromeda Galaxy). 20 2-second exposures with Opticstar CCD on 14-inch w/ focal reducer. Fully processed. This galaxy is 2.5 million light years distant!!

To get rid of the hot pixels, you use a "dark frame". This is simply an image with no illumination (again, the scope is covered) taken under the same circumstances as your real astro image. Darks should be taken the same night the real image is taken. It's considered good practice to take about the same number of exposures and same exposure time that you have of the real image and average them into one happy dark frame.

Above: M15. 20 2-second exposures with Opticstar CCD on 14-inch w/ focal reducer. Fully processed.

To minimize vignetting, you use a "flat frame". Flat frames are images with an even illumination. You can use the sky at twilight or a white screen. The recommendation is to take several flats and combine them. The software I use (Nebulosity) scales the intensity of the flats, so the exposure time isn't so important. I use the same exposure time for the flats as I do for the real images.

The flat frame should be pre-processed, that is you should subtract the dark frame and the bias frame from the flat.

The Nebulosity software has a processing algorithm. All I have to do is combine the dark frames and bias frames, and pre-process and combine the flat frames.

Above: My flat frame. Combination of 30 2-second exposures. Telescope was pointed at a white posterboard.

In summary:

To minimize read noise, subtract the bias frame from your real image.

To remove hot pixels, subtract the dark frame from the image frame.

To minimize vignetting, divide your real image by the flat frame.

Or, in equation form:

Good = (Raw - Dark - Bias) / (Flat - Dark - Bias)

The processing equation above is applied to each of the real images. Next the images are aligned and combined. Once again, Nebulosity has an algorithm for aligning and combining. For each image you select two stars, and the software will translate and rotate each image before they are combined.

Saturday night I took dark frames, flat frames, and bias frames. I also took more exposures of each image. I think the images are looking better!