REU - Summer 2010

Daniel Kenndy

B.S. Computer Science, North Carolina State University

Markerless ego-motion tracking device

Abstract:

Consider applications for a four camera device that uses optical flow to estimate its ego motion. I also attempted to improve on the device by using vanishing points in a still image to estimate each camera's rotation and infer orientation when the device is first activated. Finally, I also helped determine whether two seperate devices were moving in similar patterns. This final project was to enable a potential gesture recognition game using several devices at once.