Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-6860

Refine simple 1D DCR correction

    Details

      Description

      DM-5695 created a functional implementation of a simple DCR correction algorithm. While it appears to successfully create template images with airmass and DCR matched to science images, it is computationally inefficient and appears to introduce new artifacts to the template image. This ticket is to enhance the simple algorithm in several ways:

      • Convert to sparse matrices where possible
      • use variance weighting of the images
      • propagate masked pixels correctly
      • Refine the algorithm to mitigate the new artifacts

        Attachments

          Activity

          Hide
          sullivan Ian Sullivan added a comment -

          The transfer matrix has been redesigned as a compact kernel for each pixel, as opposed to a gigantic matrix that would transform an entire column of pixels simultaneously. This is much faster and uses much less memory, since most pixels can share the same transfer matrix, assuming DCR does not vary across an image.

          The current code can be found here:
          https://github.com/lsst-dm/experimental_DCR

          Show
          sullivan Ian Sullivan added a comment - The transfer matrix has been redesigned as a compact kernel for each pixel, as opposed to a gigantic matrix that would transform an entire column of pixels simultaneously. This is much faster and uses much less memory, since most pixels can share the same transfer matrix, assuming DCR does not vary across an image. The current code can be found here: https://github.com/lsst-dm/experimental_DCR

            People

            • Assignee:
              sullivan Ian Sullivan
              Reporter:
              sullivan Ian Sullivan
              Watchers:
              Ian Sullivan
            • Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: