Motion Assisted Capture with Template Motions

Mark David Slater. "Motion Assisted Capture with Template Motions" Master's thesis, University of California - Santa Cruz, June 2004.

The idea for my Master's thesis revolves around using a parameterized base motion as a starting point for monocular motion capture. In English, that means we're using a single camera to video tape creatures in motion, transferring that video to the computer, tracking the creature with computer vision techniques, and then translating the 2D video motion into 3D motion for our computer generated creature. My addition to this process was the addition of a generic 3D motion that gives the computer a rough idea of what the creature is expected to do. This base motion should be simple and easy (i.e.fast) to create; it should also be general enough that it can be applied to other creatures with the same structure, even if the segment sizes are different. I believed this addition will help improve the accuracy of the capture process and reduce the amount of human labor needed to capture the motion; the former turned out to be inconclusive, but the latter was true.

My advisor for my Masters work was Jane Wilhelms.