[Octave-bug-tracker] [bug #51982] [octave forge] (image) Failing unit test for nlfilter

[Carnë Draug]

Update of bug #51982 (project octave):

                  Status:                    None => Fixed                  
             Open/Closed:                    Open => Closed                 

    _______________________________________________________

Follow-up Comment #1:

Thank you for the report.  Finally got the bottom of this.

This is caused because of what nlfilter assumes to be the center of for a
sliding window with even length which is different from imdilate. In imdilate
it is:


floor ((size (nhood) +1) /2)


while in nlfilter is:


ceil ((size (nhood) +1) /2)


So while we were using imdilate to check if nlfilter was doing what it should
with the max filter, the results would always be wrong.

The reason on why the test was almost always passing was pure stupidity on my
part.  I was using a very large image of random integers in the range [1 10]
because it makes it easier to visualize.  However, I was using a very large
window (size 3x4x7x5) so it almost always included the value 10 so the result
was almost always a large arrays of 10s.

More details in the commit message 3af35ac9ead9

    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?51982>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/