Machine Vision

cancel
Showing results for 
Search instead for 
Did you mean: 

How can my camera identify different people?

I'm building a camera which can trace moving object. How can the camera only follow the one that I want? For example, two objects moving at the same time.
Thank you.
0 Kudos
Message 1 of 5
(3,215 Views)
Y.L.

I worked on a similar project in college. Let's work on your example, you have 2 objects in your image, and you can actually identify them whenever one of them moves, you can then identify the centroid of the object or any pixel value that changed in that particular object and you "remember" this pixel value, a.k.a save it, so in your next iteration for searching the object the one that is closest to that value is the same object you locate before. That's the simplest way I can think of right now.

Good luck!

Nestor Sanchez
IMAQ/Motion Support
National Instruments
Nestor
Message 2 of 5
(3,215 Views)
Nestor,

Could you please be more specific of how i can identify the the centroid of the object (or changed pixel value) when one of the objects moves.

Thank you.
0 Kudos
Message 3 of 5
(3,215 Views)
It depends on how are you identifying that there is an object. Are you filtering the image (thresholding) or similar? or are you just looking for pixel values that changed? In the first one, IMAQ Vision has tools to identify the centroid of an object if you have isolated from the background. In the second one, if you know that a pixel has changed and that represents an object just stored this value. It depends on how are you identifying an object. Save the most centric pixel of this one and then use it for comparison.

Good luck!

Nestor
Nestor
0 Kudos
Message 4 of 5
(3,215 Views)
nestor,
I have tried the method you suggested.However, when both objects are moving on the same time, i can only get a single centroid.( which is not the centroid of both objects) Do you have any good idea? You mentioned about reading the value of a particular pixel. Could you tell me how to do that?

Thank you.
0 Kudos
Message 5 of 5
(3,215 Views)