Contents I. Introduction
Many applications require fast, dense and accurate noncontact three-dimensional (3-D) point acquisition for object shape recovery. Optical triangulation is a typical technique which achieves high-range accuracy with simple calculations, provided the objects are located within a range of up to a few meters from the 3-D scanner. Depending on the particular application, however, a tradeoff between acquisition speed and the density of acquired 3-D points (voxels) must be found. Many commercial triangulation-based systems use light sheet projection to achieve cost effective range maps acquisition. Here, a two-dimensional (2-D) position detector array like a regular imaging sensor is used to recover the profile of the intersection between the light sheet and the object being imaged. Full range maps are obtained by scanning the light sheet over the entire object. In this case, angular resolution is limited by the sensor size while depth accuracy is limited mainly by the algorithm used for centroid calculation, the pixel pitch, the pixel fill factor and speckle-induced noise. Commercial systems are limited to about 15–30-kHz 3-D acquisition rate. Custom sensors that achieve real-time range map acquisition have been reported in literature [1]–[4]. They are in a development stage and are aimed mainly to robotic guidance applications.
