|
To most humans familiar with the Roman alphabet, this is clearly the letter “D”. Some other forms of machine vision might detect this fact, but that would depend on "lucky" tuning of adjustable parameters. Many methods of machine vision consist of forms of template matching. This character makes a pretty good template match with “O”. The bumps on the left cause only about a 5.2% mismatch with it. |
More analytical methods of machine vision will define “D” the way we learn to define it in school. There, we say it is part of a circle, with a gap on the left, with the gap bridged by a vertical line. Objectively, this figure has no vertical line on the left. It contains a complete circle, with no gap for the non-existent line to bridge. The quadrupole convolution yields an unambiguous answer. |