How does the function detectMSERFeatures in matlab work?

1
votes

It is known that the Maximally Stable Extremal Regions (MSERs) can found from an image in MATLAB using detectMSERFeatures, and the outputs of this function are some MSERs.

But i have one questions which have confused me for a very long time, that is, why do some MSERs may contain some other MSERs in the output MSERs? Such as, three outputs MSERs, m1, m2, and m3 may have the relationship, m1 ∈ m2 ∈ m3

1
matlabcomputer-visionmatlab-cvstmser

1 Answers

2
votes

There is a section in the detectMSERFeatures documentation (under the More About section) that may help you better understand the algorithm:

http://www.mathworks.com/help/vision/ref/detectmserfeatures.html

You can get nested regions when you have an image intensity profile that that looks like a "V". If you use the bucket filling analogy from the documentation, then as you start to slowly pour water into the "V", stable regions will begin to appear and will be labeled as an MSER region. As you continue pouring water, larger stable regions will form and they will contain the previously found MSER regions.