Tangra 1.4


 
  TANGRA     VIDEO RENDERING     PHOTOMETRY     PRE-PROCESSING     TRACKING     LIGHT CURVES     ASTROMETRY     ADD-INS     LICENSE     DOWNLOAD  

Refining


Tracking in Tangra is the process that attempts to locate and center the apertures on the measured objects as they move around because of wind, shaking, motor drive errors, drift through etc. For an occultation light curve to be a good measurement a good tracking is required and the first step to a good tracking is a well refined position and brightness for each of the tracked objects.

Before starting the measurement Tangra runs a number of refining frames to determine better the position of the objects, the distances between them, the average brightness and the brightness fluctoations (flickering). Tangra also uses the selected objects to compute the averaged PSF model of the stars. This may be used in PSF and Optimal Extraction photometry when configured.

While the refining process is running, as indicator of the progress arcs are drawn around the objects which are completed to full circles with the completion of the refining. By default Tangra is using 16 refining frames but this value can be modified from the Tracking settings. The configured number is the number of successfully refined frames which is frames on which all objects have been located. If some of the stars are faint or there is a significant noise present in the video it may take more frames than the number of configured refining frames to complete the refining process - i.e. some frames will not be used in the refining process. In such a situation using integration is recommended.

Tangra currently supports two types of tracking - "Untracked Measurement" and "Tracked Asteroidal Occultation". As the names suggest the first type of measurement does not involve any tracking. The aperture remains on fixed X, Y coordinates and measurements are taken of whatever falls within the apperture. The "Untracked Measurement" is limited to supporting only aperture photometry but the maximum supported aperture size is fairly large - 17 pixels.

"Tracked Asteroidal Occultation" type of measurement does track on the selected objects. This measurement is optimized to work for asteroidal occultations where the total measurement duration is usually not larger than 10 min. The "Tracked Asteroidal Occultation" measurement will not work well with mutual and lunar events. New types of measurements will be added for those in future.

Tangra optimizes its tracking algorithms depending on the number of selected stars and depending on the additional flags specified by the user. Those flags are displayed at the bottom of the "Light Curve Reduction" form and include things like: "Full or almost full disappearance", "Wind, shaking or movement", "Strong flickering", "Field rotation" and "Drift through". There is also an option for Tangra to "Stop on lost tracking". All of these flags tune the tracking engine to know better what to expect from the video and what to consider as "normal" and what not. With this the tracking is improved and generates less incorrectly rejected frames. However if the flags are used incorrectly the result may be an increased number of incorrectly successfully tracked frames.

Additional tracking related settings are available on the Tracking tab of the program settings. A good idea would be to increase the number of the refining frames as this may improve the tracking. However a reason to not increase your refining frames would be if your video processing routine includes a cutting of only a smaller part of the video for processing, and this smaller chunk doesn't have sufficient number of frames before the occultation that could be used for refining. Still to increase the tracking precision you should consider changing your routine so a larger number of freames are available before the occultation. Ideally the maximum of 64 refining frames should be used.

Another option which is particularly useful for people that observe frequently in wind conditions is the "Automatic recovery from lost tracking". When Tangra is unable to locate all relevent objects on a frame it marks the frame as unsuccessfully tracked. After this if the automatic recovery is enabled then another tracking algorithm will kick in and try to identify the objects using pattern recognition. In this process Tangra will enumerate all objects on the video and try to find a pattern of triangles and polygons that match the refined positions and distances between the measured objects. This process is a lot slower than the normal tracking but gives better results in cases of more serious shaking. The recovery from lost tracking can be furhter tuned from the "Tracking Tolerance" setting. The higher the tracking tolerance the slower the recovery and the more likely the tracking to be recovered. The lower the tracking tolerance, the faster the recovery and the less likely the recovery to be successful.

The "Automatic recovery from lost tracking" is only available if more than one non comparison object is selected. For example the automatic recovery cannot work if there is one occulted star and one comparison star or if there is only one occulted star. For best results use 2 or 3 guiding stars and 1 occulted star.

In a case of an unsuccessful tracking, recovery or measurement Tangra will display a warning icon for the relevant objects when the light curve is reviewed. The icon will additionally display as a hint a more detailed message of what has caused the tracking to fail. Here is the full list of warning and information messages that can be displayed by Tangra:

  Type Displayed Message     Type Displayed Message
Tracking Object with fixed manually positioned aperture   Tracking Tracking has failed because the brightness fluctoation was out of the expected range
Tracking Tracking has failed because the a PSF could not be fitted   Tracking Tracking has failed because the center of the PSF fit was too far from the expected location
Tracking A fully disappearing object was marked tracked without being detected (assuming it has disappeared)   Tracking No guiding stars have been reliably located, all objects are marked as suspect
Tracking The expected object position is outside the FOV   Recovering The object was recovered after a distance alignment
Recovering The object was recovered after searching in a wider area   Recovering The object was recovered after using pattern recognition
Recovering Failed to recover the object after a distance alignment   Recovering Failed to recover the object after searching in a wider area
Recovering Failed to recover the object after using pattern recognition   Measuring Aperture contains no pixels
Measuring The PSF fitting failed   Measuring The object apeared too far from the expected position
Measuring The FWHM of the PSF fit was out of the expected range        


When an object is added for measurement/tracking it can be added either as a guiding star, occulted star, manually positioned aperture or comparison star. Depending on the type of the added object the tracking will behave differently. More information about how to add a new object as one of those four type please see the topic on Photometry.