3D Markerless Motion Tracking in Real-Time Using a Single Camera

摘要

We present a novel motion tracking system that estimates the three-dimensional position of a moving object in real time by analyzing the image stream from a single lowest-end camera. Tracking is achieved without the need of any markers, calibration, or previous knowledge about the object. Our proposal can be applied given there is enough brightness contrast of the object with its surroundings. Such a technique allows for new Human-Computer Interaction solutions to be implemented in already running systems without a significative deployment expense.