Closed mawc2019 closed 1 year ago
There should definitely be a way to turn this off.
A boolean argument warn_cusp
is added. If True
, warning will be given when the input 2d image is likely to contain sharp corners or cusps; if False
, warning will not be given. The default option is False
.
In this PR, the Harris corner detector from OpenCV is used for detecting sharp corners or cusps in binary images. If a sharp corner or cusp is detected, a warning message will be printed out. The detection is not successful in every cases. A feature may be mistakenly classified as a sharp corner or cusp if the resolution is too low. On the other hand, a sharp corner or cusp may not be recognized if the angle is smaller than ~30° or larger than ~120°. Changing some parameters can easily increase the true positive rate, but also increase the false positive rate. This issue was partially investigated here.