OpenVC Blob/Object Tracking

This demo illustrates the motion/blob-tracking algorithm possible with OpenCV. The image is Gaussian-blurred, filtered, eroded, and finally dilated, at which point the second task of the algorithm is to simply find the center of the object and track it.

The filter idea and original source was written by a friend of mine at UNSW (Frank Paino), which I've since modified (enhanced the thresholding functions).

We did not use any blob-finding libraries (despite what you see being compiled at the start of the video (we've not been able to make use of that yet, but I'm still hopeful we'll figure it out).

Also, the following of the blob at the moment is very primitive as it simply averages the min/max of all blobs on the screen - which is ok as this has merely been a proof-of-concept. Rewriting that function is next and if people are interested, I'll upload the follow up videos.

Category:

Science & Technology

Tags:

• opencv
• blob-detection
• motion-tracking
• video-processing
• machine-learning
• artificial-intelligence

[MrSundaymonday]

Can we detect multiple objects simultaneously with technique???? Which techniques of object seg and tracking have you used?? Does it work fine in real time environment?? Pls may I get the source code?

Chandra Mani