Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-9168

Provide HSC test data for deblending

    Details

    • Type: Story
    • Status: Done
    • Resolution: Done
    • Fix Version/s: None
    • Component/s: None
    • Labels:
      None

      Description

      Identify deblender test cases in PSF matched coadds from HSC (as provided by DM-9035) which include:

      • Line of sight projections as blends, where colour information is available;
      • Cluster cores, where colour information is not helpful;
      • Large extended galaxies with internal structure in blends;
      • Extreme colour artefacts.

        Attachments

          Activity

          Hide
          rearmstr Bob Armstrong added a comment -

          I have visually inspected the cosmos field to try to find examples of these different cases. I tried to get a mix of something of everything. Since only a few patches were run with psf-matching in DM-9035, I reran the whole cosmos field. To reduce the amount of blurring, I chose to follow the same settings as the goodSeeing version in DM-9035. The y-band has some funny features so I choose to only run griz data. Here are the visits that I used

          • HSC-g
            • 29340^11690^11692^11694^11696^11698^11700^11702^11704^11706^11708^11710
          • HSC-i
            • 30486^30488^30490^30492^30494^30496^30498^30500^30502^30504^19680^19658^19660^19662^19694^19696^19698
          • HSC-r
            • 1216^23718^23704^1202^1204^23716^23706^1214
          • HSC-z
            • 17932^17944^17946^17948^17950^17952^28354^28356^17934^17900^17962

          There are some other features in this processing:

          • The galaxies with substructure have been smoothed by the psf-matching, so we will need to see how well the deblender performs on different data. I have still left the galaxies in this sample to give some more variety.
          • Due to the psf-matching, when we merge catalogs from different bands the size of a footprint can extend to the size of the whole image. Because of this I have just extracted a fixed postage stamp around each object. So this may include objects outside of the blend.
          • The data images and catalogs can be found on tiger at Princeton in /tigress/rea3/cosmos_examples.

          Here are thumbnails of the chosen images.

          Peter Melchior,Fred Moolekamp please take a look at these and see if there other types of objects we should include in the test sample.

          Show
          rearmstr Bob Armstrong added a comment - I have visually inspected the cosmos field to try to find examples of these different cases. I tried to get a mix of something of everything. Since only a few patches were run with psf-matching in DM-9035 , I reran the whole cosmos field. To reduce the amount of blurring, I chose to follow the same settings as the goodSeeing version in DM-9035 . The y-band has some funny features so I choose to only run griz data. Here are the visits that I used HSC-g 29340^11690^11692^11694^11696^11698^11700^11702^11704^11706^11708^11710 HSC-i 30486^30488^30490^30492^30494^30496^30498^30500^30502^30504^19680^19658^19660^19662^19694^19696^19698 HSC-r 1216^23718^23704^1202^1204^23716^23706^1214 HSC-z 17932^17944^17946^17948^17950^17952^28354^28356^17934^17900^17962 There are some other features in this processing: The galaxies with substructure have been smoothed by the psf-matching, so we will need to see how well the deblender performs on different data. I have still left the galaxies in this sample to give some more variety. Due to the psf-matching, when we merge catalogs from different bands the size of a footprint can extend to the size of the whole image. Because of this I have just extracted a fixed postage stamp around each object. So this may include objects outside of the blend. The data images and catalogs can be found on tiger at Princeton in /tigress/rea3/cosmos_examples. Here are thumbnails of the chosen images. Peter Melchior , Fred Moolekamp please take a look at these and see if there other types of objects we should include in the test sample.
          Hide
          pmelchior Peter Melchior added a comment -

          These look good. Inasmuch as having artifacts in the images can look good, that is... But we asked for those as well, so I think the sample will be fine for testing under real-life conditions.

          Both of the issues with the test data, arising from the PSF matching, should not be problematic for the new deblender, as long as you include the peaks of all objects in the image.

          Show
          pmelchior Peter Melchior added a comment - These look good. Inasmuch as having artifacts in the images can look good, that is... But we asked for those as well, so I think the sample will be fine for testing under real-life conditions. Both of the issues with the test data, arising from the PSF matching, should not be problematic for the new deblender, as long as you include the peaks of all objects in the image.
          Hide
          fred3m Fred Moolekamp added a comment -

          I like the samples you have selected, they will certainly be a good test of the new deblender. I'll defer to Peter Melchior as to whether or not we need any additional types of objects.

          Show
          fred3m Fred Moolekamp added a comment - I like the samples you have selected, they will certainly be a good test of the new deblender. I'll defer to Peter Melchior as to whether or not we need any additional types of objects.
          Hide
          rearmstr Bob Armstrong added a comment -

          I have put the data at http://tigress-web.princeton.edu/~rea3/cosmos_blend_examples/. I have tried to remove any lsst dependencies, but I may have missed something.

          Show
          rearmstr Bob Armstrong added a comment - I have put the data at http://tigress-web.princeton.edu/~rea3/cosmos_blend_examples/ . I have tried to remove any lsst dependencies, but I may have missed something.
          Hide
          rearmstr Bob Armstrong added a comment -

          It seems this is a sufficient start for a set of HSC deblending examples. I am closing the ticket.

          Show
          rearmstr Bob Armstrong added a comment - It seems this is a sufficient start for a set of HSC deblending examples. I am closing the ticket.

            People

            • Assignee:
              rearmstr Bob Armstrong
              Reporter:
              swinbank John Swinbank
              Watchers:
              Bob Armstrong, Fred Moolekamp, John Swinbank, Peter Melchior
            • Votes:
              0 Vote for this issue
              Watchers:
              4 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved:

                Summary Panel