Yeah. Hubble has a couple advantages over me. Aside from a couple of hundred time more light gathering power than mine, it doesn’t have to look through atmosphere. Its sorta like trying to read a newspaper that is on the bottom of a pool, through binoculars, while standing at the edge of the 3 meter board, while some fat kid at the other end is splashing around making waves.
The processing I do is basically divided into 2 broad categories. Restoration and enhancement.
Restoration:
1. Atmospheric Dispertion- If you look a a raw frame from one of my AVI’s
http://webpages.charter.net/JasonIrby/astro_pics/2003_08_16/BestFrame.jpg you can see a red halo on the upper edge and a blue halo on the lower edge of the planetary disk. When a bright object like Mars is less than straight overhead, it is hitting the surface of the atmosohere at an angle. Essientially, its having to pass through a wedge of refractive air. That spead the light of the image out a tad like a prism spread light out into a spectrum (red at one end, blue at the other). I can fix that to a large extent by splitting my raw image into its RGB channels then aligning those channels back into registration.
2. Atmospheric Distortion- The air is constantly boiling as cells of warmer air float up from the surface of the earth like bubbles from a pan of boiling water. On some nights its less, on some nights its more. These cells of warm air have a slightly different refractive index than the surrounding air. That causes the image of Mars to shift and warp and distrort its shape. However, one of the advantages of using one of these little webcams on planets is that I can take a much high frequency of images (5,10,15,30 frames per sec) Ideally, I’d capture a 1500 frames or so per AVI. Then I can sift through the frames and find ones that by luck captured the image in between distorting waves.
http://webpages.charter.net/JasonIrby/astro_pics/2003_08_16/BestFrame.jpg is a sample of the best frame from a 250 frame AVI,
http://webpages.charter.net/JasonIrby/astro_pics/2003_08_16/WorstFrame.jpg is the worst frame from that same dataset.
3. Signal to Noise Ratio- One of the drawbacks to taking a bunch of very quick frames is that each frame tends to contain a lot of noise. The main problem with this is that it limits the degree I can enhance the real image without also enhancing the noise. However, by taking say the best 200 frames out of 1500 and averaging their corresponding pixel values (after careful alignment), the noise, which is random in nature, should tend to cancel out and the signal (the true image underneath) should tend to be strengthened. The result should be an image that I can enhance much more aggressively before artifacts due to noise reaches an unacceptable level.
4. Color balance- Hey, your eyes can play tricks with you at 3am.
The eye is much less sensitive to red light. Also I have all the windows styles on my laptop set to red so the image of mars didn’t look nearly so er…pinkish at the time. But color balance is really the least of my worries. That can easily be re-adjusted later when I’m re-combining the aligned RGB channels. In this dataset, I just needed to increase the relative strength of the green channel to get back to a hue closer to what I was seeing in the eyepiece.
5. There are all kinds of fancy restoration algorithms for sharpening an image. Some of them work better on stellar subjects than extended objects like a planet surface. Examples:
http://www.mlunsold.com/examples/party.html ,
http://www.mlunsold.com/examples/comparele.html . I had less than expected benefit from these. I think there were two main causes. I was only working with about 250 frames instead of 1000-1500. I couldn’t get enough good frames to stack to get my S/N ratio high enough to support these operations without creating artifacts. Also, I believe my focus was off. For some reason these algorithms work better on blurring due to atmosphere better than due to poor focus. There realy is no substitute for perfect focus. For my data, just a couple of iterations of mild unsharp masking worked as good as anything. If I had better raw frames, thing might be different.
Enhancement:
6. My enhancements mainly involves contrast streching, non-linear histrogram modification to tease out the most detail. I consider this enhancement instead of restoration because I’m generally increasing the constrast beyond what I was seeing at the eyepiece.
Wab
