Fix aruco _filterTooCloseCandidates

[spirraw]

_filterTooCloseCandidates was occasionally not choosing the biggest
contour of a candidate group.

Firstly, if the biggest contour of a candidate group happened to be
the first candidate of that group, it would never be chosen due to the
comparison loop skipping it.

Secondly, contour sizes were being compared by their circumference
(via counting the number of points in a contour) rather than area.
In some cases it is possible for a smaller candidate to have more
points than a bigger one. This was fixed by comparing the contour
area as defined by the contour's corners.

Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

• I agree to contribute to the project under Apache 2 License.
• To the best of my knowledge, the proposed patch is not based on a code under GPL or other license that is incompatible with OpenCV
• The PR is proposed to proper branch
• There is reference to original bug report and related work
• There is accuracy test, performance test and test data in opencv_extra repository, if applicable
  Patch to opencv_extra has the same branch name.
• The feature is well documented and sample code can be built with the project CMake

force_builders=linux,docs

[alalek]

@spirraw Thank you for contribution!
Could you please attach an image (lossless .png format is preferable) with this problem? So we can validate the fix.

/cc @szk1509 Could you please take a look?

[spirraw]

Here is an image with which the issue should be reproducible.

Here is a visualization of the contours that are passed to _filterTooCloseCandidates:

They are grouped correctly, however, because the inner candidate has more points than the outer one, it is incorrectly chosen as the biggest contour in the group.

[Chris-Bee]

This might be related to the incorrect settings of the threshold windows described here #2811 . Using the settings described there might solver your issue.

[spirraw]

Prior to making this patch, I was also able to work around the problem by adjusting the adaptive thresholding parameters. Doing so eliminates the inner contours entirely, and as a result the bugs in _filterTooCloseCandidates are mitigated. While it is possible to to work around specific cases by adjusting the adaptive thresholding parameters, doing so may adversely affect performance in other cases. This patch is about fixing two underlying issues in _filterTooCloseCandidates, the function's entire purpose is to filter out these "inner contours".

[szk1509]

hi @spirraw :)
it all looks pretty reasonable, I haven't had the time to do the testing I normally do, but in broad strokes a big 👍

[alalek]

Thank you 👍