Abstract: vision system described in this paper has been developed to calculate the 3-D surface patches of objects illuminated by a projected grid. In order to exploit the parametrization offered by the projected grid, we have to extract the imaged grid as a network of curves, rather than a graph, so the 2D and 3D processings are based on the curve idea. One difficulty in this approach is to establish the correspondence between the original grid and the imaged grid. Different from other published methods, the correspondence is done curve per curve without any ambiguity by using geometrical and global constraints. The 3D projected grid is calculated per curves by triangulation, to obtain 2 independent families of 3D curves allowing to calculate the shape parameters.!10
展开▼