Pinpoint centroids

Discussion in 'Legacy Models - Community Support' started by Don, Apr 16, 2016.

  1. Don

    Don Cyanogen Customer

    Joined:
    Nov 16, 2015
    Messages:
    30
    Does Pinpoint, the full version, use the centroids determined by the same algorithm used by Aperture mode in the Information window? Or, when Pinpoint is activated, does it use an entirely new technique in Astrometric mode, and re-scan the image? It appears that the centroid values in PinpointLE are the same, but I would like to know if this is also true for the full version. - Thanks!
     
  2. Bob Denny

    Bob Denny Cyanogen Customer

    Joined:
    Oct 12, 2014
    Messages:
    1,115
    Location:
    DC-3 Dreams, SP, Mesa, Arizona +1 480 396 9700
    PinPoint LE and PinPoint full use exactly the same centroiding algorithms (and there is a choice in PP full). The aperture tool in MaxIm is completely separate.

    But why do you ask? Is there an overriding question you have?
     
  3. Don

    Don Cyanogen Customer

    Joined:
    Nov 16, 2015
    Messages:
    30
    I am trying to compare Astrometrica with Pinpoint on a sample image, to see which program I should use. When I pick a star in MaximDL and use the aperture tool, I get a certain centroid. When I then solve the image using PinpointLE, and look at the astrometry tab, I get the same centroid, to 0.001 pixels. Unfortunately, I tried to evaluate Pinpoint-full using MaximDL, but my camera was delayed by two months, so my trial ran out just when I started the analysis. All I have for now is PinpointLE. If you can reset my trial, maybe that would help.

    I am trying to get just accurate centroids, never mind matching with stars, for now. When I compare centroids with Astrometrica and MaximDL on the same image, I get differences in the 0.1 pixel range for stars with SNR greater than about 20. Astrometry.net and Prism are worse, so I've narrowed my choice down to Astrometrica and Pinpoint.

    Once I get the most consistent centroids (my FWHM are only between 1.0 and 1.4 pixels, so the gaussian curve fitting is difficult for any program), chosen by comparing the differences between several images, then I'll start the star-matching. Maybe I need to average the centroids from the two programs, if the answers are effectively random at the 0.02 pixel level.

    I understand that Astrometry.net uses a quadratic fit in two dimensions to get the centroid - that is why it is not very accurate for my particular case. Unfortunately, I don't know much about the details of the other programs. I really like the way that MaximDL centroids are calculated; the result is practically independent of the aperture center, to +/- 0.001 pixel/pixel, usually. If Pinpoint uses a similar algorithm, that would be a good choice. I am looking at as many tests as I can in the next few weeks.

    Thanks,
    Don
     
  4. Bob Denny

    Bob Denny Cyanogen Customer

    Joined:
    Oct 12, 2014
    Messages:
    1,115
    Location:
    DC-3 Dreams, SP, Mesa, Arizona +1 480 396 9700
    Well I'll leave it to you to do your independent testing. As far as "star matching" it's automatic in PinPoint. But wait, there's more... PinPoint does automatic 4th order 2-dimensional distortion mapping so the transformation between XY and Equatorial is optimized throughout the field even in the presence of significant pincushion, sin(x)/x, etc optical distortion, over and above the planar projection (typically TAN). Combined with UCAC4 you will get as good results as you can get anywhere.

    But why do you need hyper-accurate astrometry? I'm not saying you are off the deep end I am just curious.
     
  5. Don

    Don Cyanogen Customer

    Joined:
    Nov 16, 2015
    Messages:
    30
    Hi,

    I am planning an experiment to repeat the experiment that made Einstein famous: measuring the solar deflection of stars during the 2017 eclipse. See my upcoming article in Sky&Tel (August 2016 issue). It's a very difficult experiment, and I need to measure individual stars on individual frames to better than 0.1 arcsec. After lots of averaging, I hope to get down to 0.02 arcsec. I'll also be talking at the SAS conference in June in Ontario, and would be happy to talk with you there. I am giving credit in all my talks and papers to the vendors, since one of the key features is how modern amateur technology can allow a relatively easy replication of this famous experiment.

    None of the current star catalogs are good enough, not even URAT1. Fortunately, ESA Gaia data is supposed to be released this fall, with micro-arcsecond accuracy combined with Tycho-2 data to improve proper motion. I also need to correct for differential refraction across my field, and have incorporated that and chromatic dispersion into the USNO NOVAS program. USNO is also advising me on incorporating pixel phase errors. Its a lot of detail, but I've done some test exposures under similar bright skies. I expect to get 8 stars during totality - not enough for an automatic program. I'll be taking calibration images during part of totality, with maybe 30 measurable stars, so I can get a last-second calibration on the plate scale and optical distortion axis. Once I get the centroids, I'll add the star positions in Excel, and do a least-squares analysis using the deflection constant as one of the variables.

    As I've said, I've used MaximDL manually to select some stars in a single image, and the PinpointLE astrometry tab in MaximDLV6 reports the same centroid numbers. What would really save me time and analysis work, is to know if Pinpoint-full uses those centroids found by MaximDL, or if you use a different algorithm to find all new centroids. I'd like to compare the centroids found by MaximDL with the centroids found by Pinpoint, if they are different. Since my FWHM are so small, the specific algorithm makes a difference. If the Pinpoint and MaximDLV6 algorithms are the same, then I can continue my analysis with PinpointLE until I get more calibration data and decide if I need to purchase your full program. Since my trial run expired before I got my camera, can I send one image to you for analysis, or can you tell me if you use MaximDL's centroids?

    Thanks,
    Don
     
  6. Bob Denny

    Bob Denny Cyanogen Customer

    Joined:
    Oct 12, 2014
    Messages:
    1,115
    Location:
    DC-3 Dreams, SP, Mesa, Arizona +1 480 396 9700
    We discussed this extensively on a phone call.
     
  7. Don

    Don Cyanogen Customer

    Joined:
    Nov 16, 2015
    Messages:
    30
    I am just starting to use PinPoint (full). I solved some of my images, and found a .stars file in the folder with the original stars. The file columns have no headers, so I don't know what they mean. It looks like the first two columns are the X and Y of the "Image Star", but I don't see a column with the measured centroids or SNR or FWHM or RA or DEC or Star ID. Is there some other file with this information?
     
  8. Bob Denny

    Bob Denny Cyanogen Customer

    Joined:
    Oct 12, 2014
    Messages:
    1,115
    Location:
    DC-3 Dreams, SP, Mesa, Arizona +1 480 396 9700
    Please post to the DC3 Dreams Comm Center at http://forums.dc3.com/ with the login you received with your purchase. This has nothing to do with MaxIm DL.
     
    Doug likes this.

Share This Page