Improvement of the spatial resolution of an optical 3-D measurement procedure

Three-dimensional (3-D) surface coordinates can be calculated by stereoscopic methods from the images produced by two CCD cameras. Classical procedures based on matching algorithms cause systematic errors at highly curved surfaces or at steps in surfaces. An alternative procedure is proposed in this paper. Instead of correlating areas in two images, grey-scale values of single pixels obtained from a number of different images taken in succession are correlated. Thus, it is possible to achieve higher spatial resolution and to reduce systematic errors at "problem surfaces".