search

UWCubeSat / lost

Lost: Open-source Star Tracker
37 stars 5 forks source link

Quest vs. DQM disparty, other issues #124

Open zeddie888 opened 10 months ago

zeddie888 commented 10 months ago

oresat-coarse

A number of slightly sus issues arise when testing on this image

  1. Quest vs DQM disparity

On this image, with a combination of COG+Tetra, using Quest vs. DQM yields pretty different results, with DQM being much closer to Astrometry's result

Astrometry: RA=76.847, DEC=46.34

Quest: RA=76.5133, DEC=38.9359, Roll=257.884

DQM: RA=76.8836, DEC=47.1353, Roll=78.0816

  1. Centroiding with COG detects over 800 centroids
  2. Pyramid performs weirdly on this image - it fails to identify a pattern using the 4 brightest stars (i.e. it will fail to identify anything when --centroid-filter-brightest 4 \ is used)
markasoftware commented 10 months ago

See issue #110 for the QUEST disparity (not sure what the cause is. But I remember from evaluations that when there is a disparity, DQM is correct, at least in all cases I saw)

Pyramid issue is worrying, I'll look into it when I have time (this weekend?). Try increasing or decreasing the angular tolerance. If this is a generated image, also try running pyramid without the centroiding step. Another possibility is that there are two separate 4-star patterns in the catalog with very similar inter-star distances; in this case Pyramid will refuse to identify them because it's ambiguous.