Three-dimensional scanner based on stereovision techniques

This work presents a three-dimensional scanner system based on stereovision techniques. The stereo-matching algorithm developed is based on multi-resolution image correlation. A new triangulation algorithm is proposed, based on the fact we digitize continuous objects. To obtain the complete description of an object, an iterative closest point algorithm has been implemented in order to register the collection of data sets into a common object-centered coordinate system. Finally, as an application of the 3D-digitization process the reconstructed model has been used to generate Fourier-space filters of an arbitrary 2D view. These filters have been evaluated by a simulation of a VanderLugt architecture.