Grooper 21.00.0082 is available as of 12-12-2023! Check the  Downloads Discussion  for the release notes and to get the latest version.
Grooper 23.1.0016 is available as of 03-15-2024! Check the  Downloads Discussion  for the release notes and to get the latest version.
Grooper 23.00.0042 is available as of 03-22-2024! Check the Downloads Discussion for the release notes and to get the latest version.

Question: Not understanding how OMR works

ptomkoptomko Posts: 68 ✭✭
edited April 2018 in The Astronauts (Q&A)
When I set up OMR with Registration and test it from the "Data Element Profiles" tab under the DocType, it looks like it is working, As see in "OMR Question.png". However, when I then go and test it from the data field, it doesn't find the Checkmark, and has several pink error boxes which I am not sure of the meaning. See "OMR Question 2.png". Can someone provide me some guidance about what I may be doing wrong here?
Tagged:

Comments

  • [Deleted User][Deleted User] Posts: 0 admin
    It looks like your second image didn't upload.
  • ptomkoptomko Posts: 68 ✭✭
    Looks like I forgot to attach the second image.
  • ptomkoptomko Posts: 68 ✭✭
    For anyone following this thread, I found that when you have a check one situation, when it finds the best match, it does so based on the fill percentage. It was suggested that I try to get as close in around the boxes as possible (another image I have has boxes). The problem with that is that you could end up with boxes of slightly different sizes. With the amount of black dots already contained by the boxes themselves, this could drastically throw off the percentages.
    Rather than try to get as close as possible to the boxes, one should ensure that the boxes are exactly the same size. This will give you the most consistent pixel percentage measurements.
    I would also recommend to actually draw the OMR zones INSIDE the boxes, so that only the marks themselves are considered and not the pixels making up the boxes.
    I think a good future enhancement would be for OMR to use line recognition to remove the boxes themselves so that only fill pixels are counted. Sometimes the "boxes" are circles, hexagons or other strange shapes, which complicates the algorithm. But I think we could easily tackle box removal.

Sign In or Register to comment.