0

I came across MSER(Maximally stable external regions) in openCV. The docs for this is not exactly very informative (here) and I understand that the algorithm for MSER on color and grayscale images is different.

Can anyone tell me if there's an advantage to using it on grayscale instad of color (other than speed) or vice versa?

Also, would appreciate an explanation of the delta, variation, diversity and evolution parameters.

Arki99
  • 256
  • 1
  • 4
  • 11
  • Possible duplicate of https://stackoverflow.com/questions/17647500/exact-meaning-of-the-parameters-given-to-initialize-mser-in-opencv-2-4-x – Rakshith G B Dec 11 '18 at 06:57

1 Answers1

0

Essentially, grey image MSER algorithm in OpenCV is from a newer publication(David Nistér and Henrik Stewénius. Linear time maximally stable extremal regions. In Computer Vision–ECCV 2008, pages 183–196. Springer, 2008.) than color one.

Color MSER algorithm runs on O(n log (log n)), which requires a bin sort of the pixel intensities at the first place. The advantage of later algorithm is that it is faster as it can run in quasi-linear-time but worst case it can run as O(n). Given that image color, to calculate intensity from RGB image will require a pixel by pixel calculation regardless, which will introduce O(n) complexity to start with. Then, Nister's algorithm(later one) will lose its advantage, at least I believe, OpenCV developers thought this way and decided to have separate algorithms for these two cases for the sake of performance.

Semih Korkmaz
  • 1,125
  • 13
  • 26