2

I want to evaluate different interest point detectors like FAST, MSER, DOH and DOG for images of certain kinds of objects.

As a first criterion I came up with repeatability under certain conditions like lighting or scale. I think it is also interesting to see which kind of region around a point favors a selection as interest point.

What are some more criteria to consider for that kind of evaluation?

Hans Sperker
  • 1,347
  • 2
  • 15
  • 37

1 Answers1

1

You probably want to look at these two papers to get a good sense of thorough evaluation:

carlosdc
  • 12,022
  • 4
  • 45
  • 62