Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-3953

Enhance pixel flags measurement algorithm

    XMLWordPrintable

Details

    • Improvement
    • Status: To Do
    • Resolution: Unresolved
    • None
    • meas_base
    • None
    • Data Release Production

    Description

      The goal of the pixel flags measurement algorithm is to reduce the image mask information around a source. The current algorithm does this on two scales: the 3x3 pixels around the peak, and the entire source footprint. I suggest that these are merely the extrema of interesting scales, and that the algorithm should be updated to record all mask bits on additional length scales.

      As an example of why this might be necessary, consider the CLIPPED mask bit, which is set when pixels are rejected from one of the input images to a coadd, signalling that the CoaddPsf may be bad at that point. Simply detecting it within the source footprint does not mean that my source may be adversely affected (as the footprint may extend beyond the important pixels of an object), but not detecting it within the 3x3 around the peak doesn't mean that my source has not been adversely affected (as the source may be extended).

      I suggest recording the bits present within each of the following scales:

      • 3x3 around the peak (as currently);
      • pixels within a footprint containing a large fraction (e.g., 95%) of the flux of the PSF;
      • pixels within a fixed aperture (e.g., 2 arcsec); and
      • pixels within the source footprint (as currently).

      Attachments

        Issue Links

          Activity

            price, jbosch — after 3.5 years, we've done nothing about this. Upon consideration, do you regard it as:

            • Urgent (ie, we should schedule it in an upcoming sprint);
            • Low priority (ie, we should leave it on the backlog for a rainy day);
            • Not useful (ie, we should just close it as “Won't Fix”)?
            swinbank John Swinbank added a comment - price , jbosch — after 3.5 years, we've done nothing about this. Upon consideration, do you regard it as: Urgent (ie, we should schedule it in an upcoming sprint); Low priority (ie, we should leave it on the backlog for a rainy day); Not useful (ie, we should just close it as “Won't Fix”)?
            jbosch Jim Bosch added a comment -

            I'd call it low priority.

            jbosch Jim Bosch added a comment - I'd call it low priority.

            Done; thanks.

            swinbank John Swinbank added a comment - Done; thanks.
            Parejkoj John Parejko added a comment -

            In our debugging of AP source detections, the opposite of these requests would be very helpful: flag sources that with the majority/all of pixels in e.g. the 3x3 center having a bit set.

            Parejkoj John Parejko added a comment - In our debugging of AP source detections, the opposite of these requests would be very helpful: flag sources that with the majority/all of pixels in e.g. the 3x3 center having a bit set.

            People

              Unassigned Unassigned
              price Paul Price
              Jim Bosch, John Parejko, Nate Lust, Paul Price, Robert Lupton
              Votes:
              0 Vote for this issue
              Watchers:
              5 Start watching this issue

              Dates

                Created:
                Updated:

                Jenkins

                  No builds found.