Tracking a pedestrian in a real world image

I have a set of color, real world images that contain each contain a pedestrian. The pedestrian location is randomized for each showing - meaning that the pedestrian for image A will be in a different location each time it is viewed. How can I track the pedestrain's location as it changes location on the image?

Answer 1

[Image Analyst]

Do you have the street scene with no pedestrians at all in it? Then you might be able to subtract, unless there is big lighting changes also in the image, and cars.

Have you seen Steve's blog for today: http://blogs.mathworks.com/steve/2012/11/05/computer-vision-and-image-processing-in-r2012b/ By coincidence, it's exactly on this - pedestrian tracking.

[Sara]

I do not. The scenes have a pedestrian in them, with the x,y location known, but then the scene is scrambled and the pedestrian is relocated. I'm trying to figure out how I can determine it's new location.

[Image Analyst]

And what do you have to say about Steve's blog on pedestrian tracking?

[Sara]

helpful! I'm working with the people detector system now.

[Sara]

Do I need to download to Vision toolbox for the people detector system to work?

[Image Analyst]

Yes.