Ticket #771 (closed enhancement: fixed)

Opened 10 years ago

Last modified 10 years ago

Masks needed for "good" and "bad" measurement flags

Reported by: ktl Owned by: rhl
Priority: normal Milestone:
Component: meas_algorithms Keywords:
Cc: Blocked By:
Blocking: Project: LSST
Version Number:
How to repeat:

not applicable

Description

lsst::measure::algorithms::Flags should define masks that indicate which measurement flags indicate that a measurement is questionable (e.g. saturated, close to an edge, difficult to measure a shape) and which measurement flags are "normal" (e.g. object detected in an unbinned image).

Without such masks, it will be difficult for clients of measurement such as WCS determination to be able to filter measurements reliably as new flags are added.

Change History

comment:1 Changed 10 years ago by rhl

  • Status changed from new to assigned

I'm not sure that this is really possible, as there's a large gray area where an object is good for something but not for something else --- a user cannot avoid learning what these flags mean, and which are relevant for their problem.

We could define a "bad" mask that most users would agree on, I suppose. I don't see the use for a "good" mask --- no-one should ever say if s.mask == 0x0:; rather they should always be testing specific bits, or maybe a bad mask: if not (s.mask & BAD):

comment:2 Changed 10 years ago by rhl

  • Status changed from assigned to closed
  • Resolution set to fixed
  • reviewstatus changed from notReady to selfReviewed

I added Flags::BAD in r12004 (measAlgorithms.Flags.BAD from python).

Note: depending on your objectives, object that are not labelled BAD may need to be excluded from your analysis.

Note: See TracTickets for help on using tickets.