Has anyone used MSER in OpenCV to discover regions?

Question

Looking through this article for Reliable Text Detection , which talks about using MSER (Maximum Stable External Regions) to localize regions in an image, since it is generally more robust than the usual component analysis we have done so far.

MSER is mentioned in section 2.1 of the text detection document (if you really checked it: D), from where I cannot figure out how to apply / implement this in OpenCV. The documentation does not contain any examples, so its a little difficult to run any code.

Has anyone tried this before?

The code from the implementation is there , however there are some color conversions and stuff that is a bit confusing. Can someone help me figure out what exactly is done in MSER, and how to do it from the point of view of OpenCV? Thanks in advance.

EDIT: Found MATLAB implementation here .

Answer 1

I frankly recommend the VLFeat MSER implementation you found. I have used it in the past for a good effect.

Note that the VLFeat implementation has C / C ++ for basic numbers with C and the Matlab API. This allows you to quickly prototype using Matlab, and then easily switch to C.

For experiments with detection / description of functions, I personally prefer this library for OpenCV.

Answer 2

OpenCV already has an MSER implementation from version 2.1

Here is the C ++ documentation. The C function is cvMSER, I think, but it may be wrong, have not used this for a while.