Position recognition with Kinect

edit

Hi, I’m working on body position recognition using Kinect and openni_tracker, specifically the origin points (x, y and z) of the transformations (head, torso, elbow, hand…) given by the tracker. I have the points and now I pretend to work with them to recognize some specific positions of the body previously memorized.

I was told that Mahalanobis distance should work fine with this variables but I want to know if there is some other way to do it, some algorithm or a program already done.

I’m using ROS Groovy, so the MIT package is not an option, since I couldn’t make it work.

Thanks in advance.