3d tracking + image subtraction = specific blob detection


Just a little experiment that I performed at my company at DSign labs. Thought of sharing it with you guys.

Dhruv Adhia

Hey Dhruv

Nice. PointGrey Bumblebee?

I was doing something similar here, but using the brightest pixels from the depth map to create an image to feed into the blob tracker.


I look forward to the Natal/Kinetic camera coming out for a low cost way to do this better with nicer depth image.

Here is the same concept with edge detection

[quote author=“Dhruv”]Here is the same concept with edge detection

Nice! So what will happen if you use no holes at findcontours… I mean could be able to detect the siluet such as normal bkground substraction?