APPARATUS AND METHOD FOR INDICATING A TARGET BY IMAGE PROCESSING WITHOUT THREE-DIMENSIONAL MODELING

摘要

Using a pair of cameras, the coordinates of a target on a plane to which a user is pointing may be obtained without three-dimensional modeling and using only data derived from the respective images and no three-dimensional scene data. Each camera views at least four registration points on the plane and an indicator of a direction along which the target lies. A linear transform of a first image maps the planar projection of the direction indication into the second image. The coordinates, in the second image, of the target are determined from the intersection of the projection of the direction in the second image and the transformed projection from the first image. In another embodiment, the directions are mapped to a third reference frame or image by respective linear transforms. An application of the system would allow a user to indicate a location on a projection or television screen using a static pointing gesture. No information about the locations of the cameras is required, so the system can be set up quickly.