NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: Photographic lunars
From: Greg Rudzinski
Date: 2009 Dec 12, 12:18 -0800
From: Greg Rudzinski
Date: 2009 Dec 12, 12:18 -0800
Peter, Can you tell us more about sub pixel processing? I'm not so sure sub pixel processing is a very good way to add precision to angular measurements on a digital image. Better to stick with whole pixels. Experiment with partial field neutral density filtering of the Moon when bracketing exposures. Give black and white a shot to improve sharpness. Greg On Dec 11, 11:31�pm, NavL...@fer3.com wrote: > Hi all, > > I've recently become interested in celestial navigation and > thought I'd try a lunar distance measurement via digital camera. > It seems to work. �After skimming the interesting archives of this > group, perhaps it's worth sharing the workflow for this. > > I'm using a Canon 40D with an 85mm lens at f/4 and ISO 800. �No > single exposure time gives both lots of stars and a well-defined > lunar limb, so the camera was set to exposure-bracket by +- 2 > stops (which is just barely enough). �The two exposures at 1/80 > sec and 1/5 sec were used and the center one at 1/20 sec discarded > (being the worst of both worlds). �All images were taken on a > tripod less than a second apart in time (by virtue of the > automatic bracketing), so the camera pointing should change very > little, limited only by the stiffness of the tripod and camera > body under the stress of the shutter operation. �Image scale is > about 20 arcseconds per pixel when considering only the green > pixels (raw images were used and the G pixels extracted from the > Bayer matrix, discarding R and B). > > So the image at 1/80 sec shows a considerably overexposed Moon > (with nevertheless a pretty sharp limb) and no stars, and the 1/5 > sec shows a number of stars in the field of view with a hopeless > blob of an overexposed moon at the center. > > Having a little experience with the free CCD astrometry pipeline > from Astromatic (http://www.astromatic.net/) and the cool work at > astrometry.net, I thought the easiest way to reduce the images > would be not to try for any "lunar distances" but just to go > directly for a lunar position in global coordinates. > > astrometry.net can take any image and find out where it's pointed > in the sky with no prior information whatsoever. �Quite amazing. > Feeding it the 1/5 sec image results in a world coordinate system > for the whole image, mapping (x,y) pixel coordinates onto > (ra,dec). �It also estimates lens distortion, which is about 1% > in this case. �About 50 stars are detected with good coverage over > the whole field (except near the moon). > > Then it's a matter of using the 1/80 sec image to estimate the > center of the moon in pixel coordinates, then using the other > image to translate that to the moon's right ascension and > declination. > > The results are: > > estimated moon center from images: > � ra 2h 31' 7", dec 19d 46' 42" > > moon's position at my location using planetarium program (Stellarium): > � ra 2h 31' 18.0", dec 19d 46' 40.8" > > Woohoo---11 arcseconds error. �With well-exposed stellar fields > I usually get errors around 2 arcseconds rms (~0.1 pixel) against > good star catalogs like UCAC-3, so perhaps most of that 11 > arcseconds is the error in estimating the subpixel position of the > lunar limb (which could maybe be improved with better image > processing). > > Now I'm sure one objection to all this is that it requires a > stable platform. �But with further playing around, maybe some > handheld images would be usable if the short exposures were used > to derive a "track" of the pointing instability using the sharp > lunar images. �Could take a stream of, say, 100 images over the > course of a minute or so, locate the images with the smallest > image-to-image movement, then look for (possibly somewhat > streaked) star images in the interspersed longer exposures. > > I'd be glad to upload the images if there's interest. > > Cheers, > Peter Monta > > ------------------------------------------- > [Sent from archive by: pmonta-AT-gmail.com] -- NavList message boards: www.fer3.com/arc Or post by email to: NavList@fer3.com To , email NavList+@fer3.com