| 摘要 |
Apparatus for deriving a reconstruction (30) of an object (3) from first and second images comprises a camera (1) movable freely between two viewpoints (1, 1'). The camera orientations are arranged to be parallel at the instant of capture of the images, eg by superimposing the camera image plane at the second viewpoint on the camera image plane at the first viewpoint and tilting the camera about the X, Y and Z axes until the lines (L1 to L3) connecting correlated points of the first image and the new image seen by the camera converge at a common "vanishing point" (VP). The line (ST) joining the camera positions is derived and a partial 3D reconstruction is derived by aligned virtual projectors (PR1, PR2/PR2') located an arbitrary distance apart (in dependence on a parameter entered by the user) on this line. The process is repeated from other pairs of viewpoints and the resulting partial 3D reconstructions scaled and combined on screen. |
