CODEWITHSUNDEEP
c++opencvimage-processingcomputer-vision
Jompa234
Jompa234

Reputation: 1248

Libraries or algorithms for blob detection in stereo images

I am looking for a way to detect blobs in stereo-images but have not been able to find anything yet. I want to detect objects like for example people standing in front of the camera and cut that object out to continue processing on the blob without the background. Does anyone know about any libraries or algorithms for blob detection / tracking in stereo images?

Upvotes: 0

Views: 893

Answers (3)

rliberal
rliberal

Reputation: 53

If you have static objects you can try with this BGS method.

https://www.science.uva.nl/research/isla/pub/Zivkovic04icpr.pdf

This BGS was implemented in this library, take a look and select your best BGS.

https://code.google.com/p/bgslibrary/

Upvotes: 0

Max Allan
Max Allan

Reputation: 2395

If you can expect your background to be static relative to the foreground you could try the OpenCV background subtractor. There are two versions based on modelling each pixel with a Mixture of Gaussians, constructing a background model from the distributions which persist longest (i.e. they are static). Pixels can then be labeled as foreground when their values shift beyond a threshold from the learned background distribution.

Upvotes: 2

Wayne Uroda
Wayne Uroda

Reputation: 5095

Are you trying to segment the image based on depth data (is this what you mean by a blob?) Here is a dumb idea - what if you smooth the left and right images and then compute the difference of the smoothed images - if you do this you might be able to get an object in the foreground to stand out (at least, you might be able to get its outline). It probably depends on your foreground/background distances and the optics.

Upvotes: 0

Related Questions