We recently presented a talk at the Embedded Vision Summit in Boston, speaking about two key computer vision techniques: feature detection and tracking. In this talk, we give an overview of where feature detection and tracking is used, and explain in more detail how the Harris corner detection algorithm works and is optimized for our parallel video processor. We then continue to give a high-level overview of the Lucas Kanade feature tracking algorithm. The Embedded Vision Alliance hosted this event, and recorded the session.
The full 22-minute presentation can be viewed after registering for free at the Embedded Vision Alliance.