Kinect Skeletal Tracking algorithm - basis

Question

I'm searching information how skeletal tracking algorithm is working on Microsoft Kinect. I need theoretical basis, not examples how to implement skeleton tracking using some SDK. Does anyone has information about it. Does it similar for all SDK's (Microsoft SDK, OpenNI, OpenKinect) or it depends on them. I will be very grateful for any informations.

Answer 1

This pdf tries to explain the way kinect sensor works to recognize gestures and skeleton

http://pages.cs.wisc.edu/~ahmad/kinect.pdf

Hope it helps

Answer 2

For the original paper, supplemental materials and any other details about skeletal tracking algorithm, you can visit this page on Microsoft Research.

For a quick "how it works" about the depth sensing technology of Kinect (idea and patent by PrimeSense), I suggest you to see two videos:

1. How the Kinect Depth Sensor Works in 2 Minutes, which should help you to understand the principle of structured lights, on which the depth sensing technology is based
2. How Microsoft Kinect works with Infrared (to see AFTER the previous video), in which you can see how the infrared pattern is actually projected in the scene by the depth sensor