search

smroid / cedar-solve

A fast lost-in-space plate solver for star trackers.
https://tetra3.readthedocs.io/en/latest/
Apache License 2.0
1 stars 0 forks source link

improve performance at small 'pattern_checking_stars' values #2

Closed smroid closed 5 months ago

smroid commented 5 months ago

The default value of 'pattern_checking_stars' is 8. When solving 11deg FOV images with 20-30 detected stars against a mag=7 database, I find that solving frequently fails. Increasing 'pattern_checking_stars' to e.g. 12 helps in the few cases I've investigated so far.

Because 'pattern_checking_stars' uses the N brightest detected stars, 8 should have been a reasonable default-- we need more than 4 to allow bright interlopers (e.g. planets) to be skipped, and to deal with brightness variations due to variable stars, different spectral response of the camera vs. the star catalog, etc.

This issue is to track my investigation (and hopefully improvements) around why pattern_checking_stars=8 is often insufficient.

smroid commented 5 months ago

The 'pattern_checking_stars' parameter is no longer meaningful. At solve time, potentially all 4-star pattern combinations are formed from the N centroids; the patterns are processed in descending order of brightness. Patterns are tried until a match occurs, or there are no more patterns, or the solve_timeout is reached.