Object detection question

[speaker]

Hello. Happy New Year.

I am trying to write a Matlab program which will find
objects as they are placed on a counter top. (For example,
if a backpack is placed on a countertop, the program should
recognize the location and boundaries of the backpack.)

Right now, I keep track of an average background, then use a
background subtraction algorithm to find the new objects.
How do I look at a background-subtracted image and find the
boundaries of the new object? Is there a function in Matlab
which will do this?

The problem is that the new objects appear "patchy" in the
background-subtracted image -- some parts of the object are
light, while other parts are dark, particularly if the
colors of the new object are similar to the colors of the
average background.

Does anybody have expertise in this area?

Thanks!

[shwaip]

could you post an example of one of the background subtracted image? It may be as simple as using a simple edge-detector filter, but I'd need to see the pictures.

[shwaip]

and, i suppose an example of an image when there's no object to detect would be useful.