Has anyone developed an algorithm for determining the minimum enclosing circle for objects in a photo of the same hue (+/- a small variation of hue)?
This is for detecting red sprites in a photograph. If the minimum circle is quite small compared to the dimensions of the photo, there is a good chance of detection.
Thank you for the comment. After doing more reading, it appears that cvblob or similar libraries are actually what is needed. I played around yesterday with findContours() and MinCircle() and they are not what is needed, the cvblob class seems to be the way to go. I am new at opencv and still learning the code (and classes) as you can tell.
what does that do that OpenCV itself doesn’t? the only source for “cvblob” I found looks to be dead since 2012, and its copy on github hasn’t moved either, if you discount some changes to some prose.
With a slight bit of egg on my face, I will have to study OpenCV: cv::SimpleBlobDetector Class Reference a bit more. I meant that looking at the overall view, blob detection with a hue seemed to be the fundamental requirement. Thank you for steering me back to opencv.
SimpleBlobDetector is quite a class. you’ve got a ton of filtering options. some of them are on by default, with some default values.
you’d best disable all the filters explicitly, and for any that you want to try, enable and parameterize them explicitly. that way you’ll see what each option does.