0

I would like to detect salient regions in a color image by using Maximum stable extremal region (MSER) algorithm as described in paper http://users.cecs.anu.edu.au/~roland/Courses/ENGN8530_CVIU/donoser_bischof_CVPR2006_Efficient_MSER_Tracking.pdf

Currently iam using OpenCV with MSER Algorithm to get the Keypoints, but what i really want is a binary Image. How iam able to get this binary Image with regions of interest? Is it possible with OpenCV or must i reimplement the MSER algorithm?

shexbeer
  • 3
  • 2

1 Answers1

0

The result of the MSER is points, not the contours, so you can draw the all points on a black image to get the binary image.

Refer the code from github https://github.com/Palethorn/Feature-extraction/blob/5ad0abfb25dc7cf80de1e9083cc955ebb78c8af0/blob_detectors/mser.cpp

beenfrog
  • 66
  • 3
  • thank helped a lot, i was confused because i used detect method which fills only a vector of Keypoints. then i used the operator() which fills a 2 dimensional vector of Points which then could be plotted on a black image to reveal the mser's instead of keypoints as features – shexbeer May 21 '14 at 12:09