Frame differencing with Kinect and opencv


Is there any way to do frame differencing with openCV and/or the depth data of Kinect to detect interaction in a scene?

In the scene there are no skeletons, just a table with some objects on it and I need to detect when any of the objects are added/removed!

This is part of my thesis, I’d really appreciate any kind of help!

Yes, it is possible. You can install the OpenCV and boofCV libraries and explore the provided examples. You can even do this with plain Processing Java and raw pixel manipulation. Explore those examples and put together a demo sketch to test your idea. I recommend you work first with images (video stream) from the camera. If you want to work with the depth, you can do that at a later stage when you need to put some effort to fuse your camera and depth data.