3D Input Using Hand-held Objects and Computer Vision

Abstract

An untethered hand-held object whose pose can be determined automatically is useful for human-computer interaction (HCI) applications such as gaming and augmented reality (AR). Computer vision and ducial marker systems can be employed to turn ergonomic objects such as rubber balls, toy swords and guns, balloons, etc. into low cost input devices. A digital video camera such as a consumer webcam is all that is needed to allow the 3D pose of a naturally held object to be used as input to control a CAD system, game, or other GUI interface. Markers are placed on the object whose positions are learned by the system automatically and used to determine object pose in a real time system