On Wed, 25 Feb 2009 07:54:20 -0800 (PST), clay@claysturner.com wrote:>Hello All, > >In the world of astronomy we currently have comet Lulin which will be >observable in binoculars and telescopes for another couple of months. >So I set up my scope to take a picture of it. So the DSP part comes in >two ways. > >First since the Earth rotates you need to have your scope track the >stars to counter the rotation. Even with a moderate quality "clock >drive" on a scope, the tracking is not generally great and when you >try to track for a minute or more errors (periodic and misalignment) >in the clock drive will show up. So the fix these days is to piggyback >a smaller scope onto the main scope and put a ccd camera on the "guide >scope." The ccd camera is connected to a laptop via usb and the camera >is read into the computer every few seconds. You select a star that >the camera sees to use as a guide star and the computer tracks the >errors in the measured star position when compared to the ideal and >these errors are then Kalman filtered and the corrections are sent to >the telescope's mount to keep the guide star centered up. > >Another DSP aspect (a pretty simple one yet effective) is taking >multiple short exposures (1 to 10 minutes each) and stacking them >together to average out the noise. You may also shoot a dark frame and >use that to flatten out thermal noise in the sensor. This often shows >up lighter portions along the edge or corner of the sensor. > >Here is a result: Comments welcome. > >http://www.claysturner.com/CometLulin.jpg > >I hope you all enjoyed.Nice!! Any idea what the net magnification wound up being? I suspect that the low magnification needed for many comet observations helped this turn out as well as it did. Eric Jacobsen Minister of Algorithms Abineau Communications http://www.ericjacobsen.org Blog: http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php
Slightly OT: Comet Lulin and Kalman Filtering and signal averaging
Started by ●February 25, 2009
Reply by ●February 25, 20092009-02-25
Reply by ●February 25, 20092009-02-25
clay@claysturner.com wrote:> On Feb 25, 12:05 pm, "bungalow_st...@yahoo.com" > <bungalow_st...@yahoo.com> wrote: >> On Feb 25, 10:54 am, c...@claysturner.com wrote: >> >> >> >> >> >>> Hello All, >>> In the world of astronomy we currently have comet Lulin which will be >>> observable in binoculars and telescopes for another couple of months. >>> So I set up my scope to take a picture of it. So the DSP part comes in >>> two ways. >>> First since the Earth rotates you need to have your scope track the >>> stars to counter the rotation. Even with a moderate quality "clock >>> drive" on a scope, the tracking is not generally great and when you >>> try to track for a minute or more errors (periodic and misalignment) >>> in the clock drive will show up. So the fix these days is to piggyback >>> a smaller scope onto the main scope and put a ccd camera on the "guide >>> scope." The ccd camera is connected to a laptop via usb and the camera >>> is read into the computer every few seconds. You select a star that >>> the camera sees to use as a guide star and the computer tracks the >>> errors in the measured star position when compared to the ideal and >>> these errors are then Kalman filtered and the corrections are sent to >>> the telescope's mount to keep the guide star centered up. >>> Another DSP aspect (a pretty simple one yet effective) is taking >>> multiple short exposures (1 to 10 minutes each) and stacking them >>> together to average out the noise. >> do you cross correlate each exposure to line it up before stacking?- Hide quoted text - >> >> - Show quoted text - > > I did this one manually, but I do have a program for aligning and > rotating each image. It needs some improvement. Rotation of each image > is needed if your telescope does not use a polar (equitorial mount). I > do have such a mount so translation will suffice for me. For people > using alt-azimuth mounts even if they keep a star in the field > perfectly aligned from image to image, the field will rotate about > that star. If you combined exposures add up to minutes or more this > will become apparent. > > I'm working on an auto align program to speed up this process. It is > not unusual for one to combine a hundred images! I'm still new at this > so I haven't done one with more than 10 images yet. > > ClayI have a Schmidt-Pechan prism with 1.2" clear aperture in a standard 1.25" mount that can rotate the image back, keeping it stationary. (I have no drive for it.) I don't use it much, so you can borrow it to play with if you ever get a Dob. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●February 25, 20092009-02-25
On Feb 25, 12:27�pm, Eric Jacobsen <eric.jacob...@ieee.org> wrote:> On Wed, 25 Feb 2009 07:54:20 -0800 (PST), c...@claysturner.com wrote: > >Hello All, > > >In the world of astronomy we currently have comet Lulin which will be > >observable in binoculars and telescopes for another couple of months. > >So I set up my scope to take a picture of it. So the DSP part comes in > >two ways. > > >First since the Earth rotates you need to have your scope track the > >stars to counter the rotation. Even with a moderate quality "clock > >drive" on a scope, the tracking is not generally great and when you > >try to track for a minute or more errors (periodic and misalignment) > >in the clock drive will show up. So the fix these days is to piggyback > >a smaller scope onto the main scope and put a ccd camera on the "guide > >scope." The ccd camera is connected to a laptop via usb and the camera > >is read into the computer every few seconds. You select a star that > >the camera sees to use as a guide star and the computer tracks the > >errors in the measured star position when compared to the ideal and > >these errors are then Kalman filtered and the corrections are sent to > >the telescope's mount to keep the guide star centered up. > > >Another DSP aspect (a pretty simple one yet effective) is taking > >multiple short exposures (1 to 10 minutes each) and stacking them > >together to average out the noise. You may also shoot a dark frame and > >use that to flatten out thermal noise in the sensor. This often shows > >up lighter portions along the edge or corner of the sensor. > > >Here is a result: �Comments welcome. > > >http://www.claysturner.com/CometLulin.jpg > > >I hope you all enjoyed. > > Nice!! � Any idea what the net magnification wound up being? � �I > suspect that the low magnification needed for many comet observations > helped this turn out as well as it did. > > Eric Jacobsen > Minister of Algorithms > Abineau Communicationshttp://www.ericjacobsen.org > > Blog:http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php- Hide quoted text - > > - Show quoted text -Thanks - This was shot in prime focus meaning the DSLR is in the focal plane of the main lens. The sensor is 26 by 36mm and the focal length is 1019mm so the field of view works out to be approximately 1.35 by 2.02 degrees. The system has most of the coma corrected out of it, so these numbers should be within a percent or two. And most of the error is near the edges. To put size into perspective, the full Moon varies from 0.48 up to 0.55 degrees in angular diamter. This comet's size is typical. Hale Bopp at one point subtended over 40 degrees!! In that case just mount the camera with a normal lens onto a clock drive. My 'scope is photgraphically fast - it is F/4. Clay
Reply by ●February 25, 20092009-02-25
On Feb 25, 12:30�pm, Jerry Avins <j...@ieee.org> wrote:> c...@claysturner.com wrote: > > On Feb 25, 12:05 pm, "bungalow_st...@yahoo.com" > > <bungalow_st...@yahoo.com> wrote: > >> On Feb 25, 10:54 am, c...@claysturner.com wrote: > > >>> Hello All, > >>> In the world of astronomy we currently have comet Lulin which will be > >>> observable in binoculars and telescopes for another couple of months. > >>> So I set up my scope to take a picture of it. So the DSP part comes in > >>> two ways. > >>> First since the Earth rotates you need to have your scope track the > >>> stars to counter the rotation. Even with a moderate quality "clock > >>> drive" on a scope, the tracking is not generally great and when you > >>> try to track for a minute or more errors (periodic and misalignment) > >>> in the clock drive will show up. So the fix these days is to piggyback > >>> a smaller scope onto the main scope and put a ccd camera on the "guide > >>> scope." The ccd camera is connected to a laptop via usb and the camera > >>> is read into the computer every few seconds. You select a star that > >>> the camera sees to use as a guide star and the computer tracks the > >>> errors in the measured star position when compared to the ideal and > >>> these errors are then Kalman filtered and the corrections are sent to > >>> the telescope's mount to keep the guide star centered up. > >>> Another DSP aspect (a pretty simple one yet effective) is taking > >>> multiple short exposures (1 to 10 minutes each) and stacking them > >>> together to average out the noise. > >> do you cross correlate each exposure to line it up before stacking?- Hide quoted text - > > >> - Show quoted text - > > > I did this one manually, but I do have a program for aligning and > > rotating each image. It needs some improvement. Rotation of each image > > is needed if your telescope does not use a polar (equitorial mount). I > > do have such a mount so translation will suffice for me. For people > > using alt-azimuth mounts even if they keep a star in the field > > perfectly aligned from image to image, the field will rotate about > > that star. If you combined exposures add up to minutes or more this > > will become apparent. > > > I'm working on an auto align program to speed up this process. It is > > not unusual for one to combine a hundred images! I'm still new at this > > so I haven't done one with more than 10 images yet. > > > Clay > > I have a Schmidt-Pechan prism with 1.2" clear aperture in a standard > 1.25" mount that can rotate the image back, keeping it stationary. (I > have no drive for it.) I don't use it much, so you can borrow it to play > with if you ever get a Dob. > > Jerry > -- > Engineering is the art of making what you want from things you can get. > �����������������������������������������������������������������������- Hide quoted text - > > - Show quoted text -Thanks for the offer. I'm using a schmidt - newtonian on a equitorial mount. Clay
Reply by ●February 25, 20092009-02-25
On 25 Feb, 16:54, c...@claysturner.com wrote:> Hello All, > > In the world of astronomy we currently have comet Lulin which will be > observable in binoculars and telescopes for another couple of months.Interesting. Haven't heard about that one, so either it was discovered recently (I haven't paid much attention to astro stuff recentky), or it isn't visible from 65N. Or both.> So I set up my scope to take a picture of it. So the DSP part comes in > two ways. > > First since the Earth rotates you need to have your scope track the > stars to counter the rotation. Even with a moderate quality "clock > drive" on a scope, the tracking is not generally great and when you > try to track for a minute or more errors (periodic and misalignment) > in the clock drive will show up. So the fix these days is to piggyback > a smaller scope onto the main scope and put a ccd camera on the "guide > scope." The ccd camera is connected to a laptop via usb and the camera > is read into the computer every few seconds. You select a star that > the camera sees to use as a guide star and the computer tracks the > errors in the measured star position when compared to the ideal and > these errors are then Kalman filtered and the corrections are sent to > the telescope's mount to keep the guide star centered up.So you use the Kalman filter to control the mount. I have come across similar tasks in image-to-image feature tracking. Is it the same Kalman filter? If so, do you know of intro material?> Another DSP aspect (a pretty simple one yet effective) is taking > multiple short exposures (1 to 10 minutes each) and stacking them > together to average out the noise. You may also shoot a dark frame and > use that to flatten out thermal noise in the sensor. This often shows > up lighter portions along the edge or corner of the sensor. > > Here is a result: �Comments welcome. > > http://www.claysturner.com/CometLulin.jpg > > I hope you all enjoyed.I sure did. Images from the web is just about the only chance to see astro objects these days; cloud covers have been dense, thick and low for more than a month now... Rune
Reply by ●February 25, 20092009-02-25
On Feb 25, 12:21�pm, c...@claysturner.com wrote:> On Feb 25, 12:05�pm, "bungalow_st...@yahoo.com" > > > > > > <bungalow_st...@yahoo.com> wrote: > > On Feb 25, 10:54�am, c...@claysturner.com wrote: > > > > Hello All, > > > > In the world of astronomy we currently have comet Lulin which will be > > > observable in binoculars and telescopes for another couple of months. > > > So I set up my scope to take a picture of it. So the DSP part comes in > > > two ways. > > > > First since the Earth rotates you need to have your scope track the > > > stars to counter the rotation. Even with a moderate quality "clock > > > drive" on a scope, the tracking is not generally great and when you > > > try to track for a minute or more errors (periodic and misalignment) > > > in the clock drive will show up. So the fix these days is to piggyback > > > a smaller scope onto the main scope and put a ccd camera on the "guide > > > scope." The ccd camera is connected to a laptop via usb and the camera > > > is read into the computer every few seconds. You select a star that > > > the camera sees to use as a guide star and the computer tracks the > > > errors in the measured star position when compared to the ideal and > > > these errors are then Kalman filtered and the corrections are sent to > > > the telescope's mount to keep the guide star centered up. > > > > Another DSP aspect (a pretty simple one yet effective) is taking > > > multiple short exposures (1 to 10 minutes each) and stacking them > > > together to average out the noise. > > > do you cross correlate each exposure to line it up before stacking?- Hide quoted text - > > > - Show quoted text - > > I did this one manually, but I do have a program for aligning and > rotating each image. It needs some improvement. Rotation of each image > is needed if your telescope does not use a polar (equitorial mount). I > do have such a mount so translation will suffice for me. For people > using alt-azimuth mounts even if they keep a star in the field > perfectly aligned from image to image, the field will rotate about > that star. If you combined exposures add up to minutes or more this > will become apparent. > > I'm working on an auto align program to speed up this process. It is > not unusual for one to combine a hundred images! I'm still new at this > so I haven't done one with more than 10 images yet. > > Clay- Hide quoted text - > > - Show quoted text -I see, i don't work with CCD cameras, but a friend of mine works here http://www.flicamera.com/ fun stuff!
Reply by ●February 25, 20092009-02-25
On Feb 25, 10:54 am, c...@claysturner.com wrote:> Hello All, > > In the world of astronomy we currently have comet Lulin which will be > observable in binoculars and telescopes for another couple of months. > So I set up my scope to take a picture of it. So the DSP part comes in > two ways. > > First since the Earth rotates you need to have your scope track the > stars to counter the rotation. Even with a moderate quality "clock > drive" on a scope, the tracking is not generally great and when you > try to track for a minute or more errors (periodic and misalignment) > in the clock drive will show up. So the fix these days is to piggyback > a smaller scope onto the main scope and put a ccd camera on the "guide > scope." The ccd camera is connected to a laptop via usb and the camera > is read into the computer every few seconds. You select a star that > the camera sees to use as a guide star and the computer tracks the > errors in the measured star position when compared to the ideal and > these errors are then Kalman filtered and the corrections are sent to > the telescope's mount to keep the guide star centered up. > > Another DSP aspect (a pretty simple one yet effective) is taking > multiple short exposures (1 to 10 minutes each) and stacking them > together to average out the noise. You may also shoot a dark frame and > use that to flatten out thermal noise in the sensor. This often shows > up lighter portions along the edge or corner of the sensor. > > Here is a result: Comments welcome. > > http://www.claysturner.com/CometLulin.jpg > > I hope you all enjoyed.Nice work, Clay! I would like to add a shameless advertisement of a colleague's work that is similar to this: http://www.commsp.ee.ic.ac.uk/~pld/ In particular, he used multiple snapshots of a "simple" object such as the moon and combined them to extract a higher-resolution version of the object. http://www.commsp.ee.ic.ac.uk/%7Elbaboula/videos/Moon4SR.html A similar idea can be applied to interpolation of multiview images. http://www.commsp.ee.ic.ac.uk/%7Ejberent/research.html
Reply by ●February 25, 20092009-02-25
clay@claysturner.com wrote:> Hello All, > > In the world of astronomy we currently have comet Lulin which will be > observable in binoculars and telescopes for another couple of months. > So I set up my scope to take a picture of it. So the DSP part comes in > two ways. > > First since the Earth rotates you need to have your scope track the > stars to counter the rotation. Even with a moderate quality "clock > drive" on a scope, the tracking is not generally great and when you > try to track for a minute or more errors (periodic and misalignment) > in the clock drive will show up. So the fix these days is to piggyback > a smaller scope onto the main scope and put a ccd camera on the "guide > scope." The ccd camera is connected to a laptop via usb and the camera > is read into the computer every few seconds. You select a star that > the camera sees to use as a guide star and the computer tracks the > errors in the measured star position when compared to the ideal and > these errors are then Kalman filtered and the corrections are sent to > the telescope's mount to keep the guide star centered up. > > Another DSP aspect (a pretty simple one yet effective) is taking > multiple short exposures (1 to 10 minutes each) and stacking them > together to average out the noise. You may also shoot a dark frame and > use that to flatten out thermal noise in the sensor. This often shows > up lighter portions along the edge or corner of the sensor. > > Here is a result: Comments welcome. > > http://www.claysturner.com/CometLulin.jpg > > I hope you all enjoyed.Clay, You might want to see what Keith Larson is up to these days. http://home.comcast.net/~klarsondsp/ Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●February 25, 20092009-02-25
On Feb 25, 2:47�pm, julius <juli...@gmail.com> wrote:> On Feb 25, 10:54 am, c...@claysturner.com wrote: > > > > > > > Hello All, > > > In the world of astronomy we currently have comet Lulin which will be > > observable in binoculars and telescopes for another couple of months. > > So I set up my scope to take a picture of it. So the DSP part comes in > > two ways. > > > First since the Earth rotates you need to have your scope track the > > stars to counter the rotation. Even with a moderate quality "clock > > drive" on a scope, the tracking is not generally great and when you > > try to track for a minute or more errors (periodic and misalignment) > > in the clock drive will show up. So the fix these days is to piggyback > > a smaller scope onto the main scope and put a ccd camera on the "guide > > scope." The ccd camera is connected to a laptop via usb and the camera > > is read into the computer every few seconds. You select a star that > > the camera sees to use as a guide star and the computer tracks the > > errors in the measured star position when compared to the ideal and > > these errors are then Kalman filtered and the corrections are sent to > > the telescope's mount to keep the guide star centered up. > > > Another DSP aspect (a pretty simple one yet effective) is taking > > multiple short exposures (1 to 10 minutes each) and stacking them > > together to average out the noise. You may also shoot a dark frame and > > use that to flatten out thermal noise in the sensor. This often shows > > up lighter portions along the edge or corner of the sensor. > > > Here is a result: �Comments welcome. > > >http://www.claysturner.com/CometLulin.jpg > > > I hope you all enjoyed. > > Nice work, Clay! �I would like to add a shameless advertisement > of a colleague's work that is similar to this: > > http://www.commsp.ee.ic.ac.uk/~pld/ > > In particular, he used multiple snapshots of a "simple" object > such as the moon and combined them to extract a higher-resolution > version of the object. > > http://www.commsp.ee.ic.ac.uk/%7Elbaboula/videos/Moon4SR.html > > A similar idea can be applied to interpolation of multiview images. > > http://www.commsp.ee.ic.ac.uk/%7Ejberent/research.html- Hide quoted text - > > - Show quoted text -Hello Julius, Yes actually this is being done quite often nowadays. There are a few guys that are shooting with CCD video cams 1000s of images of something like Jupiter and then combining them to get amazing results. Even though the Moon keeps the same face towards the Earth, it does have libration so over time we can actually photograph 59% of its surface. I've seen these combined to produce a 3-D image of the Moon. Not that is really cool. I'm just getting started with this stuff. It is like being a kid again. Clay
Reply by ●February 25, 20092009-02-25
On Feb 25, 12:53�pm, Rune Allnor <all...@tele.ntnu.no> wrote:> On 25 Feb, 16:54, c...@claysturner.com wrote: > > > Hello All, > > > In the world of astronomy we currently have comet Lulin which will be > > observable in binoculars and telescopes for another couple of months. > > Interesting. Haven't heard about that one, so either it was > discovered recently (I haven't paid much attention to astro > stuff recentky), or it isn't visible from 65N. Or both. > > > So I set up my scope to take a picture of it. So the DSP part comes in > > two ways. > > > First since the Earth rotates you need to have your scope track the > > stars to counter the rotation. Even with a moderate quality "clock > > drive" on a scope, the tracking is not generally great and when you > > try to track for a minute or more errors (periodic and misalignment) > > in the clock drive will show up. So the fix these days is to piggyback > > a smaller scope onto the main scope and put a ccd camera on the "guide > > scope." The ccd camera is connected to a laptop via usb and the camera > > is read into the computer every few seconds. You select a star that > > the camera sees to use as a guide star and the computer tracks the > > errors in the measured star position when compared to the ideal and > > these errors are then Kalman filtered and the corrections are sent to > > the telescope's mount to keep the guide star centered up. > > So you use the Kalman filter to control the mount. I have come > across similar tasks in image-to-image feature tracking. Is it > the same Kalman filter? If so, do you know of intro material? > > > Another DSP aspect (a pretty simple one yet effective) is taking > > multiple short exposures (1 to 10 minutes each) and stacking them > > together to average out the noise. You may also shoot a dark frame and > > use that to flatten out thermal noise in the sensor. This often shows > > up lighter portions along the edge or corner of the sensor. > > > Here is a result: �Comments welcome. > > >http://www.claysturner.com/CometLulin.jpg > > > I hope you all enjoyed. > > I sure did. Images from the web is just about the only chance > to see astro objects these days; cloud covers have been dense, > thick and low for more than a month now... > > RuneHello Rune, I have an old book that is on the introduction of Weiner and Kalman filtering. It has an example about tracking a moving object. When I get back to the farm, I'll get the exact title for you. The weather around here this time of year is interesting for Astro work. Last Wed. We had tornados - the closest missed me by a couple of miles. It cut a swath 1/4 mi wide by 7 miles long. Last Thurs. The weather has cleared buy we had winds of 30 mph. I used my binoculars to see the comet. Too windy for a telescope. Last Fri. Clear and cold (below freezing) but windless - I shot pics till about 3:00AM Last Sat. Clouds Last Sun. Clouds Last Mon. Clear and cold - wind!! Clouds ever since. The past closest to Earth last night. It is now in Leo. Sky and Telescope magazine's wed site has a free star map you can down load to locate the comet. For me at midnight it is SE about 50 degrees above the horizon. You should be able to see it apart from the clouds. It is faint at the naked eye limit. Quite seeable in 7x50 binoculars. Clay
