摘要 |
An arrangement for acquiring the 3D shape of an object O comprises a monocular (Figure 9) or a stereoscopic (Figure 8) digital camera optionally provided with an inertial sensor (11) which indicates the position and/or orientation of the camera(s) in relation to the object either to facilitate the derivation of 3D information from overlapping images in the former case or the combination of 3D data sets in the latter. The inertial sensor may be dispensed with (Figures 10, 13 and 14) or used only to determine relative rotation of a single camera, translation being determined by tracking a network of points a, b, c; a', b', c' in the image I. An image processor (Figures 3, 4 and 6) may utilise converging epipolar lines (EPa to EPd) which constrain the search for correlated regions of the images, or a projected fractal pattern (Figures 15 and 15A) whose correlated regions can be identified from their identical topology in both images. |