DC-3 Dreams, SP

PinPoint Astrometric Engine 5.0

Welcome to the web site for the PinPoint Astrometric Engine, a programmable engine that provides sensitive, robust, high-speed research-grade astrometric image processing for FITS files from any camera. A set of point-and-click tools (Visual PinPoint) for fast mass plate solving and survey-level asteroid and supernova hunting are included.

Typical uses include plate solving which gives precise pointing to telescope mounts, asteroid and comet searching, high-precision astrometry (detection and measurement of positions), and supernova hunting. The package comes with Visual PinPoint, a point-and-click Windows program that has automated tools for hands-off asteroid and supernova hunting with optional blinking for validation. You can use Visual PinPoint with any FITS or POSS/DSS images.

PinPoint is used as the plate solving engine in many astronomy applications out there, including Diffraction Limited's MaxIm DL and MaxPoint, Weber and Brady's FocusMax and PoleAlignMax, CCDWare's CCD Autopilot, and of course our own ACP Observatory Control Software.

Concerned about:

Click topics to reveal details.

• Accuracy? (as good as it gets)

  The numbers speak for themselves. IAU #333 uses PinPoint with an 18" f/2.8 hyperbolic telescope and a 2K by 3K detector at 1.5 arcsec/pixel. The optics produce significant astrometric distortion which PinPoint is able to automatically remove via adaptive higher-order solution. The system is operated in search mode, thus detections submitted to the MPC appear all over the detector field, not just the center. In addition, detections are usually near the noise limit and are hence subject to random variations. Here are the statistics for 2002, #333's busiest year, taken from the MPC's Residuals Statistics for Observatory Codes page. Compare with other observatories, amateur and professional, and remember that many amateur imagers are relatively narrow field and distortion-free, plus many are being used for single target astrometry near the center of the field.

  COD  Year       Tot    <1"   <2"   <3"   <4"  <=4"
  333  2002     52423 39764 11233  1237   189     0
  
       R.A.           Decl.
  +0.02 +/- 0.66  -0.02 +/- 0.63

  These numbers are as good as any observatory out there. And this is across the entire field of a very fast instrument, a worst-case scenario. Furthermore, PinPoint has been improved through several revisions since this data was taken.

• Speed? (faster than anything else out there)

  As the size of images increases, the demands on an automated astrometric engine increase steeply. Many of the available astrometry packages bog down when asked automatically match more than a dozen or so stars, or detect stars in a field containing more than 100 detectable stars. PinPoint can handle up to 15,000 field stars, and match thousands of stars, while performing a sophisticated software field-flattening.

  The object detector in PinPoint is 400% faster than SExtractor, yet the two packages detect virtually identical sets of objects on a given image, and the XY coordinates agree to less than two tenths of a pixel. This is not to say that SExtractor's positions are "better" and PinPoint only approximates them, tests have shown that PinPoint's positions are more consistent, particularly at low signal to noise ratios.

  The following times are approximate times to complete a plate solution without knowledge of the rotation angle of the field (rotation-free solution), and only approximate centerpoint coordinates and plate scales. This is a statistical process, both for the detection of the objects against background noise, and for the catalog-field star matching and astrometric solution process. Numbers and times are approximate, and are for a 1.5GHz class machine running Windows XP:

  Size	Field	Catalog	Match	Time
  512 x 512	100 stars	20 stars	15 stars	0.5 second
  1024 x 1024	1,000 stars	300 stars	250 stars	2 seconds
  3072 x 2048	10,000 stars	800 stars	700 stars	7 seconds
  4096 x 4096(*)	10,000 stars	2,000 stars	1,700 stars	30 seconds

  * for Catalina Sky Survey images, which are 2.8 degrees wide. The sensitivity of PinPoint was reduced to limit field stars to approximately 10,000.
• Sensitivity? (as good as it gets)

  PinPoint uses an enhanced thresholded peak-finder for object detection, then uses a synthetic circular-aperture photometer to find the centroid and measure the flux of every detected object in the image. Background noise statistics are calculated first, then used to set the peak-finder's threshold in 32 zones of the image. This gives PinPoint a high degree of resistance to image shading. It is quite usable on all but the dirtiest raw CCD frames, though best results are always obtained from properly calibrated images, of course.

  In the following, use the back-arrow in your browser to return here. If you'd like to see what PinPoint can do, click here to see a blinking animation. Can you find any asteroids? Now click here to see the animation of PinPoint's automated detection. Did you find it while manually blinking? Now click here to see the report (including the MPC submission) generated by PinPoint. These images are rather poor, with a wide variation in sensitivity despite being acquired with identical exposure times. This should give you at least some confidence that PinPoint can tackle difficult images.

  The best way to evaluate PinPoint's sensitivity is to download a trial package and use the Blink tool to examine some plate-solved images. You can tell which objects are detected by the ones that PinPoint will circle in yellow. Judge for yourself.

• False matching? (an issue with other automated programs!)

  There are only a few packages that do automatic image-catalog star matching. Despite requiring you to pre-set accurate values for camera rotation angle (eliminating use on alt/az mounts without a derotator) and focal length, most of these packages still generate false match solutions. One very popular package has such a high false match rate that its usefulness as an automation tool is seriously compromised.

  PinPoint virtually never generates a false match. Sites such as the Puckett Supernova Search Team and IAU Observatory #333 have plate solved many tens of thousands of images with no false detections. PinPoint can quickly find an image-catalog star match without any knowledge of the camera rotation angle, and with only a rough approximation of the focal length. It is capable of matching thousands of stars in a single image, thus eliminating the possibility of a false match on star-rich images. It also adjusts its requirements for the number of stars matched based on the number of available stars for matching. This gives it a high degree of resistance to false matches on images with varying numbers of stars, all the way down to matches with only 8 stars.