Reconstruction of volumetric 3D models

Photo-realistic 3D computer models of real objects and scenes are key components in many 3D multimedia systems. The quality of these models often has a large impact on the acceptability of applications like virtual walk-throughs (e.g., city guides or virtual museums), caves, computer games, product presentations in e-commerce, or other virtual reality systems. Although the rendering of 3D computer scenes can often be performed in real-time even on hand-held devices, the creation and design of 3D models with high quality is still time consuming and thus expensive. This has motivated the investigation of a large number of methods for the automatic acquisition of textured 3D geometry models from multiple views of an object. The large body of work devoted to this problem can basically be divided into two di erent classes of algorithms. The rst class of 3D model acquisition techniques computes depth maps from two or more views of the object. Here, depth is estimated from changes in the views caused by altering properties like position of cameras (shapefrom- stereo, shape-from-motion), focus (shape-from-focus, shape-from-defocus), or illumination (shape-from-shading, photometric stereo). For each view, a depth map can be computed that speci es for each pixel the distance of the object to the camera. Since the object is only partially represented by a single depth map due to occlusions, multiple depth maps must be registered into a single 3D surface model.