How To Create a Moon Mosaic
Creating a Moon Mosaic
using the Meade LPI and AstroArt 4.0
Article by Phil Russell
Cygnus Observatory, Alresford
First let me introduce myself. I joined the Hampshire Astronomical Group last year as I moved into top gear building my garden observatory in Alresford. This would be the realisation of a life-time goal and something I had promised to do once I retired. Well, I retired from IBM in 2002 but it took a good four years to clear the decks of long outstanding jobs such as garden landscaping, a new kitchen, house redecoration etc, etc, before I finally found the time (at last!) to focus on the design and construction of the Observatory which I finished in Feb 2007.
Since commissioning my RCX400-12" scope I have been on a voyage of discovery learning what it can do, as so much has changed since my earlier experiences with my trusty C-8 which tailed off many years ago when the pain of setting up and taking down became too great! This was before the advent of the goto telescope and CCD imaging so it has been a delight to discover what modern technology can do - although I have to say the learning curve has been steep! I've also discovered the down-side of Astronomy - sleep deprivation! I will be happy to share how I built the observatory together with the details of commissioning my scope (quite a job!) and the results I've been able to achieve in a future talk or another article in this series.
I first got the idea for creating a Moon mosaic when I discovered how easy it was to image the moon with the Meade Luna Planetary Imager (LPI). So I thought it would be an interesting experiment to take a sequence of images along the terminator and then try and stitch them together. If I did this over the period of a waxing moon it should be possible to build up an image of successive terminator mosaics for the whole moon and thereby get a detailed image of the full moon. This all sounds very simple! In fact it turns out to be a real challenge, and one that I have not been able to pull off yet. However having tried I now have a much better idea of what is necessary to pull it off and I will share these thoughts with you.
The Meade Lunar Planetary Imager (LPI)
The Meade LPI is an inexpensive CCD camera designed specifically for lunar and planetary imaging. Its about the same size as a web cam. It was introduced by Meade in 2003 for $149 and was one of Sky and Telescope's "Hot Products" for Jan 2004. It uses a CMOS colour chip and has a VGA resolution of 640 x 480 pixels. The chip size is 5.18 x 3.90 mm and the pixels are 8.0 microns square. It comes with software from Meade, called the Autostar Suite, which also includes Planetarium software, Image Processing and Telescope Control of Meade telescopes which use the #496 Autostar or Autostar II controllers. You basically plug it into an eyepiece holder, connect the usb connection to your computer where the Autostar Suite is running and off you go.
Here's a screen capture of the LPI program running while acquiring an image:
In operation it is a sophisticated piece of kit:-
Real-time display on the PC screen
This is always available as a "Live" window tag
Colour or Mono modes are available
I have not tried using colour for the Moon yet however I have used it for Jupiter and was pleasantly surprised by the image quality.
Magic Eye software-assisted focusing
Two triangles change size as you adjust the focus on the telescope - the closer to focus the larger they are. I try to maximise the value of the left hand value in the dialog (11.94 shown) as this correlates to the quality of the focus.
Automatic and manual exposure control from .001 to 15 secs
In practice once you have centered and focused your object you have to set the exposure time. You first give the LPI a hint by selecting from an object drop-down menu the object you are going to image, i.e. the moon. This results in the Gain and Offset sliders being set to default values (similar to the brightness and contrast on your tele). You then hit the "Auto Adj" button and several images of increasing exposure time are taken to determine the best one. For the moon I found this worked reasonably well however it was about 30% too high when imaging Jupiter. I now treat the auto exposure as a guide and then try to tweak it for best results.
Many different file formats can be created
Namely Bmp, Gif, Jpeg, Png, Tif, Fits & Fits3P
Automatically takes multiple exposures
When you press "Start" the LPI camera begins taking images and continues until you press "Stop". The first 10 images are used to create a composite baseline image which is then sharpened by the LPI software and used to assess the quality of subsequent images.
Automatically selects best images
The quality of an image is assessed by comparing it against the sharpened baseline image and a percentage measure of quality is determined. If the assessed quality is greater than the value you specify, the default is 50%, then the image is kept for processing in the next step. In the example shown above you will see in the status bar bottom left that 21 images have passed the 50% quality measure so far and that this has taken 21 secs and the current frame has only 1% quality. This quality value is in fact a direct readout of the current seeing quality. Its easy to see the cycles that occur in the seeing where after a (protracted!) period of poor seeing all of a sudden the quality value leaps to 100% and more for several seconds - if you are lucky many seconds!
Automatically aligns multiple images using a user defined tracking box feature and combines them into one superior image
Tracking and Combining is optional and controlled by the settings of the check boxes - "Track" and "Combine".
If "Combine" is checked each image of acceptable quality is added into the composite image which is displayed in a new window. If "Save Every Image" is checked the composite image after each update is saved otherwise only the final image is saved.
If "Combine" is not checked then each good image can be saved for future stacking with Registax or similar.
If "Track" is checked you can draw a tracking box around an object in the "live" display to identify the image feature you want to be used for tracking. Typically this is a bright spot surrounded by a dark area. For the moon I try to use a bright spot in a crater. ( In the example above, the tracking cross is over a small bright feature that is difficult to see as its obscured by the cross.) Once done you will see the tracking cross move in the image as it endevours to keep the image centered. Its like you have enabled automatic guiding at the image level. Using this technique the program can process each image and translate it as necessary in x and y so that it can be aligned for stacking. This is a really nifty feature - it not only shows directly the effects of atmospheric turbulence, it also shows the effects of the periodic errors in your telescope drive. Very interesting!
How many images should be stacked? The manual recommends 50 as the optimum number to bring out the most detail.
Creating a Mosaic of Luna images
Here are two overlapping images of an 6 day old Moon that I took using Track and Combine. They can be stitched together using the Mosaic option in Astroart.
First a few words about Astroart 4.0 from MSB software in Italy. I first heard about this when Dr. Richard Miles mentioned he used it for image processing and photometry during his recent talk at HAG's. My ears pricked up when he mentioned the price at around 120 pounds. At about half the price of Maxim DL that sounded very interesting so I purchased it.
It is a very capable program packed full of function. It has a good look and feel and I have found it fairly intuitive to use. It comes with a 140 page manual which is concise and organised for reference rather than reading. For quick start-up it includes 12 tutorials which I found very useful.
The Mosaic option works very simply and requires just one common registration point to be identified in both images. The images are then co-aligned about this point using just translation in x and y. This is unlike the co-registering or stacking of deep-sky images where it requires three points in each image to determine any rotation and scaling as well as translation.
The Astroart cursor has a nice feature which helps you to set a registration point. When the cursor moves over an area of image which looks like a star it changes from a cross to a small circle. So if one chooses a small bright feature as a registration point one can position the cursor over it and with fine adjustment get the cross to change to a circle at which point a left-button click on the mouse records the position. I found this features allows registration points to be defined quite accurately without the need to zoom-in on the image.
The Astroart Mosaic dialog is shown above. This shows that two registration points have been defined, one in each image. There are a range of options for how to handle the pixels in the overlapping region. After experimenting I found that the "Fading" option worked quite well and the two images are blended together seamlessly as shown.
Using this technique I managed to stitch together all 13 images taken along the terminator with the following result.
As I mentioned at the beginning, to pull this off for the whole Moon with my current equipment setup would be quite a challenge. The RCX400-12 has a focal length of 2438mm and this gives a FOV at the LPI chip of 7.3 x 5.5 mins - see below for the maths. This is quite small compared to the Moon's angular diameter of about 31 mins and as a result you would need to take about 8 images with a 20% overlap to cover the terminator! In fact I found I was taking more than this, more like 10-12 images. Also, as the terminator moves about 2 mins / night you would have to image it at least every other night otherwise you would run out of overlapping features.
With this current setup the LPI has a resolution of 0.68 secs / pixel and this means the resultant image of the Moon will be about 2700 x 2000 pixels - quite large! We don't need such a high a resolution to achieve our goal and it would certainly make life a lot easier if we increased the FOV of the LPI. If we settled for a final image of about 1860 x 1400 pixels this would require an LPI resolution of about 1 sec / pixel. At this resolution the FOV of the LPI would be 10 x 8 mins - much better! To cover the terminator this time would only take 5 images and you could afford to miss 2 nights before you ran out of overlapping features. To achieve this resolution one would need to use a focal length of 1750 mm which with my telescope would require it to work at f/6 so I would need to purchase a suitable focal reducer.
In summary I think the key points to bear in mind in order to do a Moon mosaic with the LPI are:-
- Use an equivalent focal length of about 1750mm to get a reasonable image scale that is not too high
- Plan to capture all the necessary images in one lunation. If different lunations are used then you might have more difficulty stitching the images together since this may involve image rotation and scaling. On adjacent lunations this problem should be minimised.
- Plan to image the terminator every night if possible but don't miss more than 2 nights otherwise it will be difficult to stitch overlapping features
- Its essential that the equipment setup is exactly the same each time so that the image scale and image rotation remain constant across all the images. This is obviously very important.
- The best time to do this project would be during the winter months when the ecliptic is high in the sky and hopefully with some fine winter evenings.
Having amassed all the images how do we combine them into one image? My current plan is to build the mosaic's for each imaged terminator in such a way that the adjacent mosaics have minimal overlap between them. This policy will minimise the overlap area which will have to be processed when they are stitched together. Then starting with the new Moon mosaic and working towards the fuller Moon ones stitched them together using the "fading" option for the overlapping areas. I don't know how well this will work, as I have not yet done it myself but I do know it will a challenge!
Given a CCD camera with a pixel array of (Px, Py) pixels and dimensions of (X, Y) mm:
The Field of View (FOV) in the focal plane of a telescope of focal length FL mm is given by:-
FOV in the X dimension = X / FL radians = X x 57.3 x 60 / FL mins
and similarly for the Y dimension.
The pixel resolution = FOV in the X dimension x 60 / Px secs / pixel