This forum is disabled, please visit https://forum.opencv.org
 | 1 | initial version |
Yes, operator() is actually be used in detect(). Note that detect() of the FeatureDetector class effectively calls detectImpl(). So, having a look at MserFeatureDetector::detectImpl() of mser.cpp, you will see how the contourpoints are translated to keypoints: basically an ellipse is fitted to the contour, the center is used as the point of the keypoint and its diameter gets its size. The keypoint orientation and response is obviously not encoded.
