I’ve got a question about the Mexican Hat. The ideal answer would be an ImageJ-Wiki-like page where it’s explained, because I haven’t found it.
What role does the radius play and how do I know which one I need?
I assume you mean an ImageJ plugin? To which plugin to you refer to?
Or this one?
Yes, sorry. It’s a plugin. I mean this one: https://imagej.nih.gov/ij/plugins/mexican-hat/index.html
It should be the radius of the convolution kernel (the laplacian of a gaussian) which in turn relates to the scale of the detected features. You probably need to experiment with various sizes. You should perhaps take a look at Marr’s theory of vision and edge detection.
Here is a good paper on this: