I’m working on a kinetic video installation and need to track the corners of a moving rectangle in a depth map from a kinect. (See http://vimeo.com/36880386 for reference.) I want to use the position of the corners of the object to map a mask to the object. (Sorta like http://vimeo.com/11026197 but with the rectangles rotating as in the depth map video referenced above.) Any suggestions on how a cv-challenged individual might accomplish this?